Octave Help Center

Learn how to build, deploy, and optimize your GTM brain.

Octave captures everything about how you go to market, who you sell to, what you solve, how you win, and turns it into strategy and action.

?

What is Octave?

The schema layer for go-to-market. Learn how Octave turns your ICP, positioning, and competitive angles into structured intelligence that every agent and workflow can read.

1

Set Up Your Workspace

Create your workspace, provide your URL, and teach Octave the foundations: company description, capabilities, and primary buyers.

2

Build Your Knowledge Base

Define your Library elements, configure Brand Voice, Writing Styles, and Smart CTAs. Everything Octave draws from when it generates.

3

Build Your Strategy

Create Motions to generate a full ICP matrix with tailored narratives at every Persona x Segment intersection. Layer themed playbooks for campaigns, triggers, and competitive angles.

4

Deploy Agents

Six agent types take your strategy into the market at scale: Prospector, Enrichment, Qualification, Sequence, Content, and Call Prep.

5

Connect Octave Everywhere

Give every tool on your desk access to Octave's brain via MCP. Claude Desktop, Claude Code, Cursor, Windsurf, and 30+ guided skills.

6

Connect Your Stack

Wire Octave into Clay, Instantly, Smartlead, and n8n. Map agent outputs to your email tools and automation workflows.

7

Make It Smarter

Octave listens to your calls, emails, and CRM, then surfaces insights and suggests Library updates. Activity Feed and GTM Explorer close the loop.

Resource Center

Rollout guides, deep dives, and customer stories to help your team get the most out of Octave.

What's New

Latest platform updates: MCP integration, agent improvements, Resource Center, Brand Voice, Smart CTAs, and more.

Home What is Octave?

What is Octave?

The schema layer for go-to-market: a structured, queryable knowledge base that gives every agent, workflow, and system real-time access to your GTM strategy.

Before Octave, GTM was decks. After Octave, GTM is code.

The Context Gap

Your CRM has a schema for customer data (accounts, contacts, deals) and every tool in your stack can read it. But there's no schema for how you go to market. Your ICPs, personas, messaging, competitive angles, and use cases live in decks, docs, and tribal knowledge. No structure. No API. No single source of truth.

This is the Context Gap: the distance between your strategy and the ability of your GTM engine to execute it. Your strategy lives in unstructured artifacts. Your execution lives in tools, agents, and workflows. There's no schema connecting them, just duct tape, copy-paste, and manual effort.

Octave is the schema layer that closes that gap. It turns your ICP, positioning, use cases, personas, and competitive angles into structured, queryable intelligence so every agent, workflow, and system in your stack can read your GTM strategy the same way they read your CRM today.

Why now? AI agents are rewriting GTM: prospecting, enrichment, sequencing, content, call prep. But agents are only as good as the context they consume. Feed them stale decks and you get generic output at scale. Feed them Octave and every agent inherits your strategy, your voice, and your competitive edge on every run.

The Four Systems

Record

System of Record

Your source of truth. CRM or data warehouse. Where customer reality lives.

Salesforce · HubSpot · Snowflake
Context

System of Context

This is Octave

Where ICPs, personas, prompts, and messaging live. The brain your agents read from.

Octave · GitHub · Notion
Orchestrate

System of Orchestration

Your data pipeline: enrichment, scoring, routing, sync. Pick one.

Clay · Freckle.io · Clodura
Activate

System of Activation

Where outreach actually happens. The backbone.

Unify · Lemlist · Smartlead · Instantly
Get Started
Set Up Your Workspace
Home Set Up Your Workspace
Build the Brain

Set Up Your Workspace

The foundation for everything. Create your workspace and teach Octave who you are, what you sell, and who you sell to.

Overview
Deep Dive
Resources

What You'll Define

Company Description
How Octave understands who you are and what you do
Capabilities
Your product's core functions and strengths
Primary Buyers
2-3 distinct personas to start
Use Cases & Value
How your product delivers value + differentiation
Continue in this section
Deep Dive

Setting up your workspace in Octave is the foundation for everything that follows. It's where you teach the system who you are, what you sell, and who you sell to. Once created, Octave will generate an initial version of your workspace so it can start reasoning about your business the same way you do.

Step by Step

  1. Create your workspace — Go to octavehq.com and sign up using your Google Workspace or Office 365 account. Enter your company URL and give your workspace a name. Octave will automatically scrape your website to create an initial model of your company. If you already have materials like positioning docs, value props, or product descriptions, you can upload them here to give the system more context.
  2. Review your company description and capabilities — Octave generates a first pass of your company description, product summary, and capabilities. Review and refine to make sure it reflects your business accurately. You can edit text directly in each field, chat with Octave to improve clarity or tone, and toggle Show Changes to compare revisions. The more specific you make this early foundation, the better Octave will perform as you expand.
  3. Define your primary buyers — Add your main Personas. Start with two or three distinct profiles that represent your key buyers or stakeholders. Keep them focused and differentiated. Choose one if you target both VP and SVP of Sales, and avoid small variations of the same persona. Distinct personas help Octave reason more clearly about your target audiences.
  4. Add your use cases and value messaging — Define your Use Cases: how your product delivers value in real-world situations. Then complete your Status Quo (what buyers do today, what problems they face, why current approaches fall short) and Differentiated Value (what makes your solution unique, how it changes the story). These sections connect your product's value to buyer motivations and pain points.
  5. Finalize your setup — Once you've refined your descriptions, products, and buyer information, confirm your updates. You'll now have a first pass of your workspace with Products, Personas, and Use Cases ready to build on. From here you can add more detail, create additional personas, and continue refining how Octave understands your business.

This foundation gives Octave the context it needs to generate strong, specific outputs across every workflow, from messaging to content creation.

Continue in this section
Resources

Resource Center

Octave's Resource Center lets you connect external content to speed up your workspace setup. Rather than manually entering everything, point Octave at your existing materials and let it learn.

What You Can Connect

Website Pages
Octave scrapes and indexes your public-facing pages, filtering out SEO clutter automatically
Google Drive
Connect live documents: positioning decks, product briefs, competitive analyses
Uploaded Files
Drop in PDFs, docs, or text files with value props, product descriptions, or sales materials
URLs
Point to specific pages: competitor sites, industry reports, product launch pages

How Resources Improve Setup

Smart content ingestion — Octave selectively pulls in only what's relevant to building and maintaining your Library. No clutter from blog archives or unrelated pages.

Live re-indexing — Connected websites and Google Drive docs support ongoing re-indexing. When your product page updates, Octave detects the change automatically.

Resource-aware Library creation — When building personas, segments, or use cases, Octave can reference connected resources or be directed to specific pages for additional context.

Pro tip: Upload your best positioning docs, value prop sheets, and competitive analyses during setup. The richer the starting material, the sharper Octave's initial model will be.

Next Section
Build Your Knowledge Base
HomeBuild Your Knowledge Base
Build the Brain

Build Your Knowledge Base

The Library is the brain's memory. Everything Octave knows about your business lives here, and everything it generates draws from it.

Overview
Deep Dive
Run with MCP
What's New

Summary

This video introduces the Octave Library as the home for your team's shared definitions. It shows why keeping all GTM assets in one place creates alignment and collaboration, gives a quick tour of the Library areas, and shares practical best practices for building and refining your Library over time.

What the Library Is

  • The central home for shared definitions.
  • Captures and evolves all messaging assets in one place.
  • Creates alignment, avoids silos, and enables collaboration across the team.

Library Areas

  • Offerings — The offerings you sell.
  • Personas — The profiles of the audiences you target.
  • Use Cases — The specific problems you solve or jobs your product enables.
  • Reference Customers — Examples of real customers who validate your story.
  • Segments — Distinct groups in your market that share common traits.
  • Competitors — The alternatives your buyers may consider.
  • Proof Points — The evidence that backs up your claims. These can be metrics, or outcomes that reinforce your value.

Best Practices: Building

It is highly context-dependent based on your market and how you go to market in motion. But some general rules of thumb:

  • Offerings — Only create one if it is a completely different offering that requires distinct use cases, personas, and messaging.
  • Personas — Capture as many profiles as you need to message in meaningfully different ways. You do not need to enumerate every role, since the system is role-aware. Instead, think about how your market splits into distinct profiles that require different messaging.
  • Use Cases — Add as many as there are ways to use your product. Each should represent a clear and distinct way your solution delivers value.
  • Reference Customers — Include as many strong references as you have. Aim to cover every persona or segment with at least one example customer, so you always have proof for each wedge of your market.
  • Segments — Define distinct groups that share traits or market characteristics, making sure they are meaningfully different from one another.
  • Competitors — Capture the main alternatives your prospects might consider, and document how you differentiate.
  • Proof Points — Anchor around your strongest, most repeatable results, whether that is numbers or qualitative outcomes.

Best Practices: Refining

  • Invest some time up front to refine the Library, but it does not need to be exhaustive.
  • Make sure everyone on the team is aligned and involved so the definitions reflect shared understanding.
  • Revisit and update as you learn from the market, as your product evolves, and as your messaging matures.
Continue in this section
Deep Dive
Library: Definitions
What the different sections of the Octave Library mean

The Library is where you define the core building blocks that power your messaging in Octave. Each element connects together to give context for playbooks, agents, and campaigns.

Offerings

Offerings represent what your company sells. They can be products or services, and they serve as the foundation for your messaging in Octave. All other elements in the Library, such as personas, use cases, and proof points, connect back to an offering. An offering can represent a software product, a service package, or any defined solution you take to market.

Personas

Personas represent the types of people or perspectives you want to speak to. They are not limited to job titles. Instead, they describe what a person values, what challenges they face, and what kind of messaging will land with them. A single persona can span multiple roles if those people share similar goals or pain points.

Use Cases

Use cases describe the functional jobs or outcomes your offering enables. Each offering will often have several use cases, depending on the customer's goal or workflow. Use cases help define why someone would buy your product and how they use it to achieve a result.

Segments

Segments represent the groups of companies or teams you want to reach. They combine both firmographic traits (like size, industry, or geography) and qualitative traits (like growth stage or go-to-market maturity). Segments are how you break up your market and think about different ways to go to market. For example, you might approach one part of your market differently from another based on how they operate, what stage they are in, or how they make purchasing decisions.

Proof Points

Proof points are aggregated outcomes or observations across your customer base. They show the measurable or qualitative impact your offering typically delivers. Examples could include increases in revenue, improvements in efficiency, or time saved. While reference customers tell one company's story, proof points summarize the broader results achieved by many customers.

Reference Customers

Reference customers are specific, named customers who illustrate success with your offering. They represent concrete examples of what happened and how they achieved results, rather than aggregate trends. Each reference customer captures the details of an individual story that can later support your proof points.

Competitors

Competitors are alternative tools, vendors, or approaches your audience might use today. Defining competitors helps clarify how you position your offering and where you have a competitive advantage.

Library: Best Practices
Creation best practices by element + global guidance

The Library is where you define the key elements that shape how your company goes to market. This article outlines best practices for creating each Library element, and global best practices around bringing in context, creating the library, updating the library, and collaborating as a team.

1. Creation Best Practices (by Element)

  • Offerings — Only create a new offering if you go to market in an entirely different way. That means a distinct motion, audience, or message. Keep offerings broad enough that each one represents a meaningful change in how you sell, not just a new feature or pricing tier.
  • Personas — You don't need to list every possible job title. Focus on the audiences you message to in a meaningfully different way. A single persona can represent multiple roles if their goals or motivations overlap.
  • Use Cases — Think of use cases as your core product stories, not a list of every benefit. Start with a few clear, high-value use cases that define what your offering helps customers accomplish.
  • Segments — Focus on breaking up your market in a way that captures meaningful nuance. You don't need a long list or highly detailed breakdowns, but define enough distinct segments that you can start tailoring messaging. Identify a few clear slices that differ in how you talk to them or how they buy.
  • Proof Points — Include proof points that are strong and representative of real customer results. Focus on the data or qualitative outcomes you want to consistently message around, and keep them grounded in evidence.
  • Reference Customers — Choose reference customers that align with your key personas or segments. They don't have to be a perfect one-to-one match, but aim to have customers that map to the unique sections of your market.
  • Competitors — Capture the competitors that actually show up in conversations or influence how you position yourself. You don't need a full list of everyone in the space. Focus on where you need to differentiate clearly and how your messaging should change when they come up.

2. Global Best Practices

Bringing in Context

Bring targeted, specific context into each element rather than trying to cover everything at once. If you're building a persona, use context and examples relevant to that persona only. If you're defining a use case, include details that help explain that use case clearly. The goal is to make each Library element concise, focused, and supported by relevant context from customer data, calls, or campaigns.

Building Your Library

Spend enough time upfront to make sure your Library accurately reflects how you understand your market today. It shouldn't take weeks, but a few hours of meaningful setup is worth the investment. The goal is to build a strong foundation that represents your best current view of your offerings, audiences, and use cases. It doesn't need to be perfect, but should be close enough that you can confidently run campaigns and learn from them.

Updating Your Library

Your Library should evolve as your product, audience, and strategy evolve. You don't need to update it daily or even weekly, but plan to make thoughtful changes as you learn from campaigns, sales conversations, or customer feedback. Expect updates roughly month to month or quarter to quarter as your GTM motion matures. Keep your Library stable enough to serve as a consistent foundation, but flexible enough to capture what's changing.

Collaboration and Time Allocation

Bring in the right people when creating or refining your Library. Product, marketing, GTM, and customer-facing teams can all provide valuable insight. It's worth spending focused time early so that what you build reflects your best understanding of your market. A strong foundation makes early campaigns more effective and learnings more valuable. There's no per-seat cost to collaborating in Octave, so involve anyone who helps shape your message.

Using the Library Across Octave
How Library elements flow into Motions and Agents

The Library is the foundation of how Octave understands your market and builds messages. Every product, persona, use case, segment, and proof point you define in the Library flows downstream into Motions and Agents to shape how Octave communicates on your behalf.

How It Flows Through Octave

Library elements feed directly into Motions. When you create a Motion, Octave uses the personas and segments linked to your offering in the Library to generate a full ICP matrix with a tailored narrative at every Persona x Segment intersection. The Library provides the raw inputs; the Motion produces the strategic output.

When you connect a Motion to an Agent, that Agent determines where the target sits in the matrix (which segment, which persona) and uses the corresponding cell narrative to generate messaging grounded in that specific ICP context.

Dynamic vs Directed Use

Default Playbook — The Motion's default playbook covers the full matrix. The Agent finds the right cell at runtime and uses that narrative. This works for broad outbound where you want full coverage without manual configuration.

Themed Playbooks — Layer Motion Playbooks (Thematic, Milestone, Account, Competitive) on subsets of the grid for campaigns with a specific angle. You can let the Agent pick the best playbook at runtime or lock it to a specific one.

Both approaches draw from the same Library and the same ICP matrix. The difference is whether the Agent uses the broad default narrative or a focused overlay.

The Bottom Line

The Library is the base layer. Motions turn that structure into strategy with full ICP coverage, and Agents turn that strategy into action.

Updating the Library with Your Existing Context
Bring your own materials into Octave to sharpen the system

You can bring your own existing materials into Octave to sharpen how the system understands your company and your go-to-market strategy. If you've already done the work in slides, briefs, or docs, you don't need to start from scratch. Instead, you can upload or link that context directly into the Library so Octave's reasoning reflects what you already know: your positioning, audience insights, proof points, and more.

This helps Octave build a more accurate "brain" around your business. The richer the context you provide, the better your Agents can reason about who you serve, what you offer, and how to communicate your value.

What to Include by Element

  • Product — Any context or documents you have about the positioning of the product as a whole.
  • Personas — For each persona, if you've defined them in any documents, include those as context. Reference the "Common Job Titles" section to see how fuzzy matching works, and tweak it for titles you want to include or exclude.
  • Use Cases — For each capability or use case, include any documents on why it exists, how it is helpful, and how it drives business outcomes.
  • Reference Customers — For each one, include as much info as possible about how you helped them specifically and what they achieved. Reference Customers refer to actual customers, not ones on the wishlist.
  • Segments — For each one, include any info on how these types of companies operate, why you care about them, etc.
Styles: Brand Voice, Writing Styles & Smart CTAs
Control how your agents sound and what they offer

Beyond Library entities, your Knowledge Base includes three style controls that shape how agents write and what they offer. These live alongside your Library and apply across Motions and Agents.

Brand Voice

Define reusable brand voices with core traits, guiding principles, tonal qualities to use and avoid, key terms, and additional context. Once created, select a brand voice on any Sequence or Content Agent and it automatically applies across all generated content.

  • Core Traits — The personality attributes that define how your brand sounds (confident, approachable, technical, etc.).
  • Guiding Principles — Rules your brand always follows (lead with value, never trash competitors, etc.).
  • Tonal Qualities — Qualities to use and avoid (conversational over formal, concise over verbose, etc.).
  • Key Terms — Brand-specific vocabulary and terminology the agent should use consistently.

Writing Styles

Writing Styles are saved versions of the formatting and copy configuration from the Sequence Agent. Instead of re-configuring sentence structure, paragraph length, punctuation preferences, and other writing mechanics every time you build a new agent, save them as a reusable Writing Style and apply it in one click. The voice defines personality, the style defines execution.

Smart CTAs

Smart CTAs let you define reusable calls-to-action so you don't have to re-prompt the Sequence Agent every time. Create a CTA once with "when to use" logic, and agents dynamically select the right one at runtime based on the recipient's profile.

  • Fixed — Always use this exact CTA. Good when you have one clear ask for the entire campaign.
  • Best Of — Pick the best CTA from a curated set you define. The agent chooses based on recipient context.
  • Auto — The agent selects from your full CTA library automatically based on persona, role, and company type.
Continue in this section
Run with MCP

Connect Your Workspace via MCP

Octave exposes an MCP server you can connect from Claude Desktop, Claude Code, Cursor, or Windsurf. Once connected, you can build and manage your entire workspace conversationally.

Full Setup Guide
Get Connected (MCP)

Build & Audit Your Library

Instead of building in the UI, you can create and refine every library element through conversation. Add articles, resources, and URLs as context, and chat directly with entities while creating or editing them.

Create Entities
"Create a persona for VP of Sales — pipeline velocity, forecast accuracy, rep productivity." Claude calls create_entity with type: persona. Works for all 7 entity types.
Refine from Real Context
"Here's a call transcript. Update my VP Sales persona to reflect what she actually cares about." Claude extracts themes and calls update_entity with natural language instructions.
Audit
"Audit my library — what's missing or weak?" Runs coverage checks, finds gaps, stale content, and duplicates. Use --fix to resolve interactively.
The LFGTM plugin for Claude Code adds 30+ specialized skills including /octave:library, /octave:audit, /octave:messaging, and more.

What's New for the Library

View all updates →
Mar 2026

Brand Voice

Define reusable brand voices with core traits, guiding principles, tonal qualities, and key terms. Apply to any agent in one click.
  • Core Traits & Guiding Principles — Define the personality and rules your brand always follows.
  • Tonal Qualities — Set qualities to use and avoid so every output sounds right.
  • Key Terms — Enforce consistent vocabulary across all agent-generated content.
  • One-click Apply — Select a brand voice on any Sequence or Content Agent.
Mar 2026

Smart CTAs

Reusable calls-to-action with three selection modes: Fixed, Best Of, and Auto. No more re-prompting for every campaign.
  • Fixed Mode — Always use one specific CTA across the campaign.
  • Best Of Mode — Agent picks the best CTA from a curated set you define.
  • Auto Mode — Agent selects from your full CTA library based on recipient profile.
  • Training Examples — Provide examples to control presentation, with optional hyperlinks and display text.
Jan 2026

Resource Center

Centralized hub for connecting websites and Google Drive. Smart content ingestion, live re-indexing, and resource-aware Library creation.
  • Centralized Resource Center — Connect Octave to websites and internal files in one hub.
  • Smart Content Ingestion — Only pulls in pages and documents relevant to your Library.
  • Live Re-indexing — Detects changes and updates in connected sources in real-time.
  • Resource-aware creation — Automatically references connected resources when building Library entities.
Next Section
Build Your Strategy
HomeBuild Your Strategy
Teach It to Think

Build Your Strategy with Motions

A Net New motion and an Upsell motion for the same product are fundamentally different conversations. The prospect who's never heard of you needs different positioning, different pain framing, and different proof than the customer who already bought in and is evaluating expansion. Motions encode that difference at the system level. Pick an offering and a motion type, and Octave generates a full ICP matrix with narratives tailored to that specific go-to-market context.

Already using Playbooks? See how Motions compare.

From Playbooks to Motions →

What You Get

When you create a Motion, Octave builds the ICP matrix from the personas and segments linked to your offering in the Library. Every cell in that matrix, every Persona x Segment intersection, gets its own narrative covering target overview, operating landscape, strategic talking points, pains, benefits, and stage-based methodology (Resonate, Elevate, Compel). Each Motion comes with a Default Playbook covering the entire grid, ready to connect to an Agent immediately. You can layer themed playbooks on top for specific campaigns, triggers, and competitive angles.

Explore Motions

Click any topic below to learn how it works.

Getting Started

Creating a Motion

Select a motion type and offering, generate your ICP matrix, and review the grid.

The Grid

ICP Matrix

Segments as columns, personas as rows. A tailored narrative at every intersection.

Inside a Cell

Cell Narratives

What each ICP cell contains and how to read, edit, and refine the narrative sections.

Layered Strategy

Motion Playbooks

Default coverage plus themed overlays for campaigns, triggers, accounts, and competitors.

Continuous Improvement

Learning Loop

Library changes and engagement signals keep your narratives current over time.

Connecting the Dots

Agents & Motions

How agents use Motions at runtime to generate messaging grounded in the right ICP cell.

Next Section
Deploy Agents
HomeBuild Your StrategyCreating a Motion
Teach It to Think

Creating a Motion

A Motion pairs one offering with one motion type. Octave handles the rest: generating a full ICP matrix with narratives at every intersection, plus a Default Playbook that covers the entire grid.

Step by Step

  1. Navigate to Motions — Open the Motions section from the main navigation.
  2. Create Motion — Click "Create Motion" to start the wizard.
  3. Select motion type — Choose Net New (acquiring new customers) or Upsell (expanding existing accounts). The type shapes how narratives are written. The buyer context for a cold prospect differs sharply from an existing customer evaluating expansion.
  4. Select offering — Pick the product or service this Motion targets. The personas and segments you've tagged to this offering in the Library are what populate the matrix.
  5. Create — Octave generates the ICP matrix. This may take a moment as it builds narratives for every Persona x Segment cell.
  6. Review the grid — Segments appear as columns and personas as rows. Each cell represents a unique ICP intersection with its own tailored narrative.
  7. Default Playbook — A Default Playbook is automatically created covering the full grid. This is your baseline coverage and can be connected to an Agent without creating custom playbooks first.

Key Things to Know

  • One Motion per combo — You create one Motion per offering + motion type combination. If you sell three products and run both Net New and Upsell for each, that's six Motions total.
  • Default Playbook — Every Motion comes with a Default Playbook that covers the full matrix. You can connect it to Agents without creating custom playbooks first.
  • Matrix depends on your Library — The grid is built from the personas and segments tagged to your offering. If you add new personas or segments to the offering later, the matrix expands to include them.

Manage Motions via MCP

With Octave connected as an MCP server, you can view, create, edit, and refine your Motions through conversation — including pulling specific cell narratives to use as messaging building blocks.

List All Motions
"Show me all the Motions in my workspace." Claude calls list_motions and returns each Motion with its type, offering, and playbook count.
Next
The ICP Matrix
Back to
Build Your Strategy
HomeBuild Your StrategyICP Matrix
Teach It to Think

The ICP Matrix

The matrix is the core of every Motion. It maps your full addressable market as a grid, with segments as columns and personas as rows, so every Persona x Segment intersection gets its own tailored narrative.

📷
Screenshot: ICP Matrix grid view

Reading the Matrix

Each column represents a segment from your Library (e.g., "Mid-Market SaaS," "Enterprise Healthcare"), and each row represents a persona (e.g., "VP of Sales," "RevOps Manager"). The cell where a persona row meets a segment column is a unique ICP intersection. Click "See Details" on any cell to open the full narrative for that intersection.

The grid is populated from the personas and segments linked to the offering you selected when creating the Motion. If your offering has 4 personas and 5 segments, you get a 4x5 matrix with 20 distinct narratives, each one tailored to the specific context of that person in that type of company.

How the Matrix Ensures Coverage

Before Motions, covering every audience you wanted to reach required building individual playbooks for every audience slice you wanted to reach. If you had five segments and four personas, full coverage meant creating and maintaining dozens of playbooks, hoping none of them went stale. The ICP matrix solves this by generating coverage for every intersection in one step. You start with full coverage by default, and you know there are no gaps.

Editing ICP Cells

Every cell in the matrix is editable. Click "See Details" to open a cell, then modify any section of the narrative. Keep in mind that cell narratives are generated from your upstream Library elements (persona descriptions, segment definitions, offering positioning), so the most effective way to improve narratives is to refine those source inputs. Direct cell edits work for quick fixes, but if auto-update is ON, the system may overwrite manual edits during the next refresh cycle.

Next
Cell Narratives
Back to
Build Your Strategy
HomeBuild Your StrategyCell Narratives
Teach It to Think

Cell Narratives

Each cell in the ICP matrix contains a structured narrative that covers who this person is, what their world looks like, how to position your offering, and how to move the conversation through awareness, urgency, and action.

Video: Cell Narratives

Narrative Sections

When you open a cell, you'll see several sections that together form the complete narrative for that Persona x Segment intersection:

  • Target ICP Overview — Who this person is, what they're dealing with, why they're a target, and what success looks like for them.
  • Operating Landscape — The realities of their daily work environment that ground the narrative in practical context.
  • Strategic Narrative — Core talking points for positioning your offering to this ICP. The foundation of what an Agent will produce.
  • Pains & Consequences — Problems this person faces and what happens if they go unresolved, and how they compound over time.
  • Benefits & Impacts — What changes when they adopt your offering: structural improvements, time reclaimed, and performance gains.
  • Methodology Stages — Three stages of buyer engagement, each with its own mindset, value propositions, and talking points:
    • Resonate — Awareness stage. Meet them where they are, name the pain they recognize.
    • Elevate — Urgency stage. Raise the stakes, connect the pain to bigger consequences.
    • Compel — Action stage. Make the case for change now, with specific proof and a clear path forward.

Editing and Refining

Cell narratives are generated from your upstream Library elements — persona descriptions, segment definitions, offering positioning, and proof points. The most effective way to improve what appears in a cell is to refine those source inputs. Cells are also directly editable when you need to make targeted adjustments. Click "See Details" on any cell, then edit any section. Consider editing when:

  • The narrative misses something specific to your market that Octave couldn't infer from the Library.
  • The emphasis is wrong: the talking points highlight a secondary concern over the primary one.
  • A recent strategic shift (new pricing, new competitor, product pivot) changes how you want to position to this ICP.

Keep in mind the auto-update interaction: if the Learning Loop is active with auto-update ON, manual edits may be refreshed during the next learning cycle. Turn auto-update OFF if you want to lock in a hand-tuned version, or leave it ON if you prefer the system to keep evolving the narrative based on engagement data.

Inspect Cells via MCP

With Octave connected as an MCP server, you can browse and inspect ICP cells through conversation.

List ICP Cells
"Show me the ICP cells for my Net New motion, filtered to the Enterprise segment." Claude calls admin_list_motion_icps with segment filters and returns matching cells with summaries.
Get Full Cell Narrative
"Pull the full narrative for VP of Sales in the Mid-Market SaaS segment." Claude calls admin_get_motion_icp and returns the complete narrative: overview, landscape, talking points, pains, benefits, and methodology stages.
Next
Motion Playbooks
Back to
Build Your Strategy
HomeBuild Your StrategyMotion Playbooks
Teach It to Think

Motion Playbooks

Every Motion starts with a Default Playbook that covers the entire ICP matrix. When you need a focused angle for a specific campaign, trigger, or competitive situation, you create a custom Motion Playbook as a themed overlay on a subset of the grid.

Video: Building Motion Playbooks

Default Playbook

The Default Playbook is auto-generated when you create a Motion. It covers the full matrix, every persona in every segment, and represents your baseline messaging for that motion type. Most users can start here and connect this directly to an Agent without creating any custom playbooks. It covers every persona in every segment and works as your starting point.

Custom Motion Playbooks

Custom Motion Playbooks are themed overlays on a subset of the grid. They exist for situations where the default narrative is good but you need a sharper angle for a specific campaign or audience. There are four types:

  • Thematic — Market trends and "why now" urgency. Use when you're running a campaign tied to a specific trend: AI adoption, regulatory changes, new compliance mandates, or anything that creates a timely reason to engage.
  • Milestone — Company events as triggers. Funding rounds, leadership changes, product launches, office expansions — moments when a company's circumstances shift and your offering becomes newly relevant.
  • Account — Named account targeting. Built for ABM campaigns or strategic accounts where you want messaging tailored to a specific company's context.
  • Competitive — Competitor displacement. Takeout campaigns, win-back plays, and head-to-head positioning against specific vendors your prospects are currently using or evaluating.

You can create multiple playbooks of the same type and mix types across a single Motion. A Motion might have the Default Playbook, two Thematic overlays for current campaigns, and a Competitive playbook for a takeout list, all drawing from the same underlying ICP matrix.

Creating a Custom Playbook

  1. Select the Motion — Choose which Motion this playbook belongs to.
  2. ICP Targeting — The full Persona x Segment grid appears. Toggle specific segments and personas on or off, or use Target All / Exclude All to select which grid coordinates this playbook covers.
  3. Narrative — Choose a playbook type (Thematic, Milestone, Account, Competitive), then describe the angle you want — or pick from suggested angles.
  4. Optional Additions — For each of Use Cases, Reference Customers, and Proof Points, choose Auto (let the system select), Manual (pick specific ones), or Off (exclude entirely).
  5. Create — Octave generates the playbook with narratives tailored to your angle for each selected grid cell.

Strategy Guidance

Start with the Default Playbook before creating custom ones. Run it through an Agent, see how the messaging lands, and add overlays only when you have a clear reason for a focused angle.

Next
Learning Loop
Back to
Build Your Strategy
HomeBuild Your StrategyLearning Loop
Teach It to Think

Learning Loop

Motion narratives aren't static. They update when you change upstream Library elements (persona descriptions, segment definitions, offering positioning), and they can also incorporate signals from real-world engagement — calls, emails, and CRM activity.

Video: Learning Loop

What Feeds the Loop

Two sources keep your Motion narratives current:

  • Library changes — When you update a persona description, refine a segment definition, or adjust your offering positioning, those changes flow downstream into the ICP cells that reference them. This is the primary lever for improving narratives.
  • Engagement signals — With CRM and call recorder integrations connected (e.g., Gong, Fathom), the system processes calls (transcripts and summaries), emails (open and reply patterns), and CRM events (pipeline changes, activity logs). Over time, it identifies which talking points resonate, which pains get traction, and where positioning is falling flat.

When the Learning Loop is active, the Motion header shows "Learning Enabled" along with an update cadence. Learning counters (Learnings, Calls, Emails, CRM Events) show how much data the system has processed.

Two modes control how learnings translate into narrative changes:

  • Auto-update ON — The system periodically refreshes ICP narratives based on what it learns. You don't need to do anything, but manual cell edits may be overwritten during the next refresh cycle.
  • Auto-update OFF — The system still collects and processes data, but your narratives stay as-is until you decide to incorporate learnings. Manual edits persist in this mode.
Next
Agents & Motions
Back to
Build Your Strategy
HomeBuild Your StrategyAgents & Motions
Teach It to Think

Connecting Motions to Agents

When you configure an Agent, the first thing you select is a Motion. The Agent uses the Motion's playbooks and ICP matrix to generate messaging shaped by the right narrative for each target.

Video: Connecting Motions to Agents

Agent Configuration

In the agent setup, you select a Motion — the specific offering + motion type combination. By default, the agent uses automatic playbook selection: it evaluates all available Motion Playbooks and picks the best fit for each target at runtime. You can also narrow this to specific playbooks or lock it to one.

Playbook Selection Modes

If you don't want the agent to auto select playbooks, you can optionally narrow which Motion Playbooks the Agent draws from. Three modes are available:

  • Auto / Best Match — The Agent evaluates all available Motion Playbooks for this Motion and picks the one that best fits the target at runtime.
  • Best Of — You select a subset of Motion Playbooks, and the Agent picks among them dynamically. Use this when you want to limit the Agent to a few specific angles (e.g., only your two active campaign playbooks) without locking it to one.
  • Fixed — You link one specific Motion Playbook, and the Agent always uses that one. Use this for single-purpose campaigns where you know exactly which angle you want applied to every target.

What Happens at Runtime

When the Agent runs against a target, it first selects the right Motion Playbook (based on your configuration: auto, best of, or fixed), then determines which ICP cell the target falls into within that playbook's grid. The cell lookup works by matching the target's segment (from company attributes) and persona (from role and seniority). The Agent uses that cell's full narrative — overview, pains, benefits, talking points, methodology stages, then generates messaging shaped by the specific ICP context of that intersection.

Next Section
Deploy Agents
Back to
Build Your Strategy
HomeDeploy Agents
Put It to Work

Deploy Agents

Agents are how Octave turns your strategy into action. Each one is built for a specific part of the GTM workflow, from finding the right people to writing the message that lands. Configure them once, run them at scale.

Choose an Agent

Click any agent to see its full walkthrough, configuration, use cases, and tips.

Discovery

Prospector Agent

Finds new leads that match your target criteria. Provide a list of companies, set a max per company, and the Agent finds leads.

Research

Enrichment Agent

Performs deep research about the person or company so your messaging has richer context.

Scoring

Qualification Agent

Scores and prioritizes leads. Qualify against company, persona, playbook, or segment.

Outbound

Sequence Agent

Creates personalized sequences for leads, outbound, inbound, or warm follow-up campaigns.

Creation

Content Agent

A flexible space to design any type of messaging or asset based on everything in your Library.

Preparation

Call Prep Agent

Gathers relevant context to prepare for meetings, person briefs, company briefs, and call scripts.

Next Section
Connect Octave Everywhere
HomeDeploy AgentsProspector
Put It to Work

Prospector

Go from a list of companies to a list of the right people. The Prospector Agent finds contacts that match your personas, and the Prospector Tool lets you search and export leads on demand.

Overview
Agent Deep Dive
Tool Deep Dive
Run with MCP

Prospector Agent

The Prospector Agent helps you go from a list of companies to a list of actual human contacts that match your personas. It uses your Library, playbooks, and persona definitions to find contacts who fit your buying profile, then returns a prioritized list with job titles, LinkedIn profiles, and persona fit scores.

It operates in two modes: Single Company Mode for known target accounts, and Lookalike Company Mode to find similar companies and surface contacts at those too.

1 credit per contact found.

Prospector Tool

The Prospector Tool is a lightweight search interface for finding leads directly. Search for people by company, title, or persona match, then export results to CSV or push them via webhook to tools like Clay.

Exporting is free. You're only charged credits when using the Agent for automated prospecting at scale.

Continue with the Prospector Agent
Agent Deep Dive
Next Agent
Enrichment Agent
Next Section
Connect Octave Everywhere

Why It's Useful

The Prospector Agent uses your Octave Library, playbooks, and personas to find contacts who fit your buying profile. Point it at a list of companies (or let it find lookalike companies automatically), tell it which personas you're looking for, and it returns a prioritized list of contacts with job titles, LinkedIn profiles, and persona fit scores.

How It Works

The Prospector Agent operates in two modes.

Single Company Mode
Find contacts at specific target accounts

You provide specific companies, and the agent finds relevant contacts based on your persona and title criteria. This is best when you already have a defined target account list and need to identify who within each company to reach out to.

Lookalike Company Mode
Find similar companies and surface contacts

You provide a seed list of ideal customers, and the agent identifies similar companies, then surfaces contacts at those companies as well. Define similarity by offering type, industry, geography, employee count, revenue range, or company stage.

Both modes use your Library playbooks and personas to guide the search. You can configure by job title (exact or fuzzy matching), apply location filters, and set quality thresholds to control which contacts surface.

Setting Up the Agent

  1. Choose a Playbook — The agent automatically adds all personas tagged to that playbook. Add additional personas if you want to cast a wider net.
  2. Define Your Targeting — Add custom job titles or use fuzzy matching to expand to similar roles. Apply location filters and quality filters (for example, exclude incomplete profiles). Specify maximum results per company.
  3. Select Search Mode — Single Company Mode for specific companies you've identified, or Lookalike Company Mode to find similar companies and surface contacts.

Credit Usage: The Prospector Agent uses one credit per contact found.

Continue with the Prospector Agent
Tool Deep Dive
Next Agent
Enrichment Agent
Next Section
Connect Octave Everywhere

Prospector Tool

The Prospector Tool is a standalone search interface for finding leads on demand. Unlike the Agent, which runs automated prospecting at scale, the Tool gives you a hands-on way to search, browse, and export contacts directly.

Smart Filtering from Your Library

The key advantage of the Prospector Tool is that it pulls targeting criteria directly from your Octave Library. Instead of manually configuring filters every time, you can search by:

  • Persona — Automatically applies the job title and seniority filters you've defined in that persona. Search for your "VP of Marketing" persona and the Tool knows exactly which titles and levels to look for.
  • Segment — Automatically applies the firmographic details you've set in that segment — industry, company size, geography, revenue range, and more.
  • Playbook — Combines both. The Tool pulls in the persona filters and the segment filters tied to that playbook, giving you the full targeting picture in one click.

This means the research and definitions you've built in your Library translate directly into lead filters, no re-entering criteria or guessing at job titles.

Export

Once you have a list of matching contacts, you can export them in two ways:

  • CSV Export — Download your list for use in any tool or spreadsheet
  • Webhook to Clay — Push results directly to Clay or other platforms via webhook for immediate activation

Pricing

Exporting is free. The Prospector Tool doesn't consume credits for searching or exporting. Credits are only used when running the Prospector Agent for automated prospecting at scale.

Tool vs. Agent: Use the Tool when you want to manually search, browse, and curate a list with your Library filters applied instantly. Use the Agent when you want Octave to automatically find persona-matched contacts across a batch of companies.

Continue with the Prospector Agent
Run with MCP
Next Agent
Enrichment Agent
Next Section
Connect Octave Everywhere

Prospecting via MCP

With Octave connected as an MCP server, you can run the Prospector Tool or find people and companies directly through conversation. Each lookup that returns results costs one credit.

Find People at a Company
"Find VPs of Marketing at Stripe who match my demand gen persona." Claude calls find_person with company and persona filters, returning matched contacts with fit scores.
Find Similar Companies
"Find companies similar to Gong — B2B SaaS, Series C+, 200-1000 employees." Claude calls find_similar_companies with your criteria and returns matching accounts.
Research a Company
"What do you know about Lattice? Pull their company profile from Octave." Claude calls find_company to retrieve firmographic data, tech stack, and other indexed signals.

Credit Usage

MCP prospecting uses the same find_person and find_company tools available in the Prospector Tool. Each result returned costs one credit.

Next Agent
Enrichment Agent
Next Section
Connect Octave Everywhere
HomeDeploy AgentsEnrichment Agent
Put It to Work

Enrichment Agent

Go from a name or company URL to a structured understanding of who they are, what they do, and how your product fits into their world. 1 credit per record.

Overview
Deep Dive
Run with MCP
What's New

The Enrich Agent turns basic lead data into meaningful context, researching and summarizing people and companies, then connecting what it learns directly back to your Octave Library so every insight reflects your positioning and messaging.

Continue with the Enrichment Agent
Deep Dive
Next Agent
Qualification Agent
Next Section
Connect Octave Everywhere

Two Agent Types

There are two enrichment agents: Enrich Company and Enrich Person. Enrich Person is a superset: you get everything the company enrichment provides, plus person-level details on top.

Enrich Company
Industry, size, focus areas, products, positioning, recent news, tech stack, and a fit analysis against your library
Enrich Person
Everything from Enrich Company, plus the person's career, expertise, responsibilities, seniority, and persona match

What It Does

Both agents perform two core jobs:

Research and Summarize — Pulls and synthesizes information about the company (industry, size, products, positioning) and, for person enrichment, the individual's career, expertise, and responsibilities.

Contextual Match — Maps what it learns to the most relevant parts of your Octave Library — showing which products, use cases, personas, playbooks, value props, proof points, and references best align with the target. It doesn't just describe who they are — it interprets them through the lens of your business.

Why It's Useful

The Enrich Agent gives you sharp, immediate research without manual digging. It goes beyond generic enrichment by linking what it finds to your unique messaging and positioning, so every insight reflects how you see the market.

You can use it to save time on initial research, to prepare data for downstream automations, or to supply focused context into other agents that need it.

What It Returns

Company Intelligence
Firmographics, products, positioning, tech stack, recent news, and operating environment
Person Intelligence
Career trajectory, expertise, responsibilities, and seniority context (Enrich Person only)
Library Match
Best-fit products, use cases, personas, playbooks, and value props from your library
Confidence Score
How well the target aligns with your ICP and chosen playbook

Example Uses

Enrich a person or company ahead of a Sequence Agent run to surface a focused insight. Pass details about someone's background and seniority as runtime context to deeply personalize the email sequence around those details.

Use it to enrich CRM or lead records with structured, contextual information tied back to your Octave Library, giving future agents and automations a richer starting point.

Credit Usage: The Enrichment Agent uses one credit per person or company enriched.

Continue with the Enrichment Agent
Run with MCP
Next Agent
Qualification Agent
Next Section
Connect Octave Everywhere

Enrichment via MCP

With Octave connected as an MCP server, you can enrich people and companies through natural conversation. Each enrichment costs one credit.

Enrich a Person
"Enrich Sarah Chen, VP of Marketing at Stripe. Her LinkedIn is linkedin.com/in/sarachen." Claude calls enrich_person with name, title, company, and LinkedIn URL. Returns a full profile with career context, expertise, and library match analysis.
Enrich a Company
"Research and enrich lattice.com — what do they do and how do we fit?" Claude calls enrich_company with the domain. Returns firmographics, products, positioning, and a detailed fit analysis against your library.
Enrich Then Sequence
"Enrich this person, then write a 3-email sequence using whatever you learn about their background." Claude chains enrich_person → run_email_agent, passing the enrichment output as runtime context for deeply personalized outreach.
Run a Saved Enrich Agent
"Run my 'Deep Research' enrichment agent on the VP of Engineering at Figma." Claude calls run_enrich_person_agent with your saved agent configuration, applying your custom enrichment instructions and playbook matching rules.

Identifiers

LinkedIn profile URL is the most reliable identifier for person enrichment, followed by email, then name + company. For companies, domain (e.g., stripe.com) is more reliable than company name.

Credit Usage

Each person or company enrichment costs one credit — whether run through the agent UI, API, or MCP.

Next Agent
Qualification Agent
Next Section
Connect Octave Everywhere

What's New for the Enrichment Agent

View all updates →
Jan 2026

Custom Tasks for Enrich, Qualify & More

Bolt on an extra output step: like a mini Content Agent inside your Enrich workflow.
  • Add a custom task — Define an additional output on top of the agent's standard jobs.
  • Same context, extra output — Inherits all the context the agent already has.
  • Flexible use cases — Generate a summary, extract a data point, or produce a talking point alongside enrichment.
HomeDeploy AgentsQualification Agent
Put It to Work

Qualification Agent

Score contacts and companies against the qualification criteria you've defined in your Library. Structured, explainable scoring with per-question reasoning. Up to 3 credits (company) or 4 credits (person) per run.

Overview
Deep Dive
Run with MCP
What's New

The Qualify Agent assesses how well a specific contact or company fits your ideal customer profile by scoring them against the qualification criteria you've defined in your Library. Instead of static rules, you get detailed, multi-dimensional scoring with the reasoning behind every answer.

Continue with the Qualification Agent
Deep Dive
Next Agent
Sequence Agent
Next Section
Connect Octave Everywhere

Two Agent Types

Like Enrichment, there are two qualification agents: Qualify Company and Qualify Person. Qualify Person is a superset: you get everything the company qualification provides, plus person-level scoring on top.

Qualify Company
Scores product fit, segment fit, and playbook fit for the target company
Qualify Person
Everything from Qualify Company, plus persona fit: how well this person matches your defined buyer personas

Why It's Useful

Traditional lead scoring relies on static rules: company size above X, job title matches Y, industry is Z. The Qualify Agent uses your offering, persona, segment, and playbook qualification frameworks to deliver context-aware scoring along with the reasoning behind it.

You can use this to prioritize leads, gate automation workflows, or decide which prospects warrant deeper engagement before deploying your Sequence or Content Agents.

What Gets Scored

You choose which dimensions to activate, from a minimum of just product fit to all available dimensions. Each activated dimension costs 1 credit.

Product Fit
Is the target company a good match for your product? (required)
Segment Fit
Does this company match your target segment?
Playbook Fit
How well does the target align with a specific playbook?
Persona Fit
How well does this person match your defined buyer personas? (Qualify Person only)

How Qualification Works

Your Library contains the qualification framework. In your Product, Persona, Segment, and Playbook entities, you define specific good-fit and bad-fit qualification questions, "Is the company in a high-growth industry?", "Does the person manage a team?", etc.

Each question has configurable settings:

Weight
High, Medium, or Low: how much this question influences the overall score
Answer Type
Yes/No: how the agent should evaluate it, with good-fit or bad-fit framing
Instant Qualifiers
Questions that immediately mark a lead as qualified or disqualified, regardless of other scores

You can edit these qualification questions over time so your scoring framework evolves as your ICP changes and as you learn what actually correlates with deals closing.

The Output

When you run the agent, you get:

Overall Score
Score out of 10 with an overall rationale explaining the assessment
Subsection Scores
Product, persona, segment, and playbook: each scored out of 10, combining into the overall
Per-Question Detail
For each qualification question: the inferred yes/no answer, reasoning behind it, and a confidence score

Setting Up the Agent

  1. Configure your qualification framework — Add qualification questions to your Product, Persona, Segment, and Playbook definitions. Set weight (High, Medium, Low) for each question. Optionally define instant qualifiers or disqualifiers.
  2. Create the agent in Octave — Navigate to Agents, select "Build New Agent", and choose Qualify Agent (Person or Company). Select which dimensions to activate: Product (required), plus any combination of Persona, Segment, or Playbook.
  3. Test with a sample — Use a known LinkedIn profile or reference a past customer to validate the scoring makes sense. Review the question answers and reasoning to ensure they align with your expectations.

Use the Output

You can extract qualification scores into your workflows in many ways:

  • Route high-scoring leads directly to reps for immediate multi-channel outreach; low-scoring leads go to automated sequences
  • Use subsection scores to customize next steps, "Product fit is 9/10 but Persona fit is 3/10" → look for new leads at that company
  • Feed qualification rationale directly to your sales team as context on why a lead scored the way it did
  • Use qualification data to inform account prioritization and territory planning

Credit Usage: 1 credit per dimension activated. Qualify Company runs up to 3 credits (Product + Segment + Playbook). Qualify Person runs up to 4 credits (Product + Segment + Playbook + Persona).

Continue with the Qualification Agent
Run with MCP
Next Agent
Sequence Agent
Next Section
Connect Octave Everywhere

Qualification via MCP

With Octave connected as an MCP server, you can qualify people and companies through natural conversation. You can use either the direct qualification tools or run your saved qualification agents.

Qualify a Person
"Is Sarah Chen, VP of Marketing at Stripe, a good fit for us? Her LinkedIn is linkedin.com/in/sarachen." Claude calls qualify_person with the LinkedIn URL. Returns an overall score, subsection scores, and per-question reasoning against your ICP criteria.
Qualify a Company
"How well does lattice.com fit our ICP?" Claude calls qualify_company with the domain. Returns product fit, segment fit, and playbook fit scores with detailed reasoning for each dimension.
Run a Saved Qualify Agent
"Run my 'Enterprise Qualifier' agent on the CTO at Figma." Claude calls run_qualify_person_agent with your saved agent, applying your custom dimension configuration and scoring framework.
Qualify Then Sequence
"Qualify this prospect. If they score above 7, write a 3-email sequence. If below, skip them." Claude chains qualify_person → run_email_agent, using the qualification score to gate whether outreach gets generated.
Qualify with Context
"Qualify this lead but factor in that they just raised a Series B and are expanding into Europe." Claude passes the additional context alongside the person data, giving the agent extra signal to inform its scoring decisions.

Credit Usage

Each qualification dimension activated costs one credit. Qualify Company runs up to 3 credits (Product + Segment + Playbook). Qualify Person runs up to 4 credits (Product + Segment + Playbook + Persona).

Next Agent
Sequence Agent
Next Section
Connect Octave Everywhere

What's New for the Qualification Agent

View all updates →
Jan 2026

Qualification Agent Updates

High-effort reasoning mode, native deep research beyond LinkedIn, and advanced scoring controls.
  • High-effort reasoning — More reasoning power for complex qualification criteria.
  • Native deep research — Search the web for signals beyond LinkedIn and the company domain.
  • Advanced scoring controls — More flexibility in how scores align with downstream workflows.
Jan 2026

Qualification Question Weights

High, Medium, Low weights on individual questions, plus instant qualifiers and disqualifiers.
  • High / Medium / Low — Control how much each question influences the overall score.
  • Instant Qualifier — Immediately mark as qualified if answered positively.
  • Instant Disqualifier — Immediately disqualify for hard deal-breakers.
Jan 2026

Custom Tasks

Bolt on an extra output step: like a mini Content Agent inside your Qualify workflow.
  • Add a custom task — Define an additional output alongside standard qualification.
  • Same context — Inherits all Library, Playbook, and runtime context.
  • Flexible — Generate summaries, extract data points, or produce talking points.
HomeDeploy AgentsSequence Agent
Put It to Work

Sequence Agent

Generate targeted email sequences at scale. Octave researches each contact, understands your positioning, and generates contextually tailored messaging.

Overview
Deep Dive
Run with MCP
What's New

The Sequence Agent is your core tool in Octave for designing email campaigns and LinkedIn outreach automations. It generates personalized copy at scale by understanding your business context, copywriting preferences, and how to frame your offer in ways that connect with each prospect.

How It Works

Instead of writing manual emails to every prospect, you use Octave as an intelligent system that researches each contact, understands your strategic positioning, and generates contextually relevant messaging tailored to who they are and why they matter to your business.

Why This Approach

Most email tools force you to choose between consistency and personalization. You either write generic templates (consistent but irrelevant) or manually customize every email (relevant but unsustainable).

The Sequence Agent bridges that gap with campaigns that feel individually crafted while remaining scalable across hundreds or thousands of prospects.

The Workflow

1. Initial Configuration
Select a test subject, sequence type, length, AI model, and baseline instructions
2. Offering & Context
Choose your product, context approach (playbook auto/manual or custom), and toggle value props
3. Writing Style
Set tone, structure, CTAs, methodology, and add training examples
4. Testing & Deployment
Run tests, iterate with "Improve Email", save presets, deploy at scale

Key Concepts

Models
Pulse (budget), Echo (balanced, recommended), Harmony (complex context)
Context Selection
Best Match (auto), Best Of (auto subset), Single Playbook (manual), or Custom Context
Writing Style
Octave Methodology, Mixed, or Custom: plus training examples to teach your voice

This isn't a template tool. You're not filling in merge fields. You're configuring strategic parameters and showing Octave examples of your voice, then letting it generate unique, personalized emails for each prospect.

Continue with the Sequence Agent
Deep Dive
Next Agent
Content Agent
Next Section
Connect Octave Everywhere
Initial Configuration
Test subject, sequence type, length, AI model, and baseline instructions

Before you configure your sequence, you need to establish the fundamentals: your test subject, sequence type, sequence length, AI model, and any baseline instructions that should apply to all emails.

Step 1: Select a Test Subject

Your test subject is the person you'll use to validate outputs as you build. Every time you change a configuration, you can run a test and see exactly how it affects the email generated for this person.

Why this matters: It lets you iterate in real time instead of guessing what changes will do.

Best practice: Choose someone who's genuinely representative of your list, ideally someone you'd actually want to reach out to. If your list is diverse, pick someone from a core persona or segment.

Step 2: Choose Your Sequence Type

This determines the framing and tone of your outreach. Octave tailors the approach based on your relationship with the prospect.

  • Cold Email Sequence — For prospects with no prior relationship. The agent assumes you're introducing yourself and your offering for the first time.
  • Warm Outbound — For prospects who have engaged with you (visited your site, downloaded content, signed up for a demo). The agent can reference that prior engagement.
  • Inbound — For prospects who reached out to you first. The tone is different — they're already interested, so you're responding to their inquiry rather than initiating.
  • Custom — For any other specific use case. You can provide custom context about the nature of the relationship.

Step 3: Set the Number of Steps

  • 3-email sequence — Quick, punchy cadence. Good for highly targeted lists or simple offers.
  • 4-email sequence — The most common. Good balance between persistence and avoiding fatigue.
  • 5-email sequence — Extended cadence. Good when you have complex messaging or lower expected response rates.

This setting determines the flow and structure. A 3-email sequence has a different rhythm than a 5-email one, and Octave adjusts accordingly.

Step 4: Select Your AI Model

Different models have different trade-offs between cost and capability. Pick the one that matches your use case.

Pulse — Best for simple, straightforward offers. Use when you have a clean value prop and don't need heavy context complexity. Lowest cost.

Echo (Recommended) — Best for most typical use cases. Handles diverse personas and contexts well without being overkill. Mid-range cost. Why it's recommended: handles most use cases well.

Harmony — Best for complex organizational context. Use when you have lots of nuance in your Library (many personas, segments, playbooks) or you're passing significant runtime context (firmographics, company signals, etc.). Highest cost.

Step 5: Agent-Level Custom Instructions (Optional)

Add instructions that apply to all emails in your sequence. Think of this as baseline guardrails for the entire campaign.

Good examples of agent-level instructions:

  • Avoid spam trigger words: "Never use these words: guarantee, limited time, exclusive offer."
  • Style preferences: "Don't use em-dashes. Use short, punchy sentences."
  • Context about your list: "These contacts visited our pricing page last week."
  • Brand voice: "We're playful and irreverent, not formal."
  • Campaign-wide constraints: "Keep every email under 100 words."

When to use agent-level instructions:

  • Rules that apply across every email in the sequence
  • General context the agent should know about the campaign
  • Baseline copywriting or brand voice guidelines
  • Information about why this list is warm, cold, or special

You can override or adjust these instructions for individual emails later, treat this as a default, not a hard rule.

Pro Tip: Don't overthink agent-level instructions on your first try. You can add more sophisticated ones as you see what Octave produces. Start simple, test, and refine.

Offering & Context
Select what you're selling and which strategic context to use

Now that your agent is configured, the next step is defining what you're selling and which strategic context from your Library to use. This tells Octave what product to focus on and how to intelligently select the right messaging strategy for each prospect.

Step 1: Select Your Offering

Your offering is the product or service you're selling: the thing you're asking prospects to pay attention to or engage with. Most companies have one offering defined in Octave unless you're a true multi-product company.

If you are multi-product, you have two options:

  • Auto Mode — Octave automatically selects the best offering for each prospect based on their profile and context. Useful if you have multiple products and want the agent to intelligently route.
  • Manual Mode — You manually select a specific product or service to ensure this Sequence Agent always sells the right one.

When to use Manual Mode: If you're running a highly focused campaign around one product, pick Manual and hard-code that offering.

Step 2: Choose Your Context Selection Approach

After selecting your offering, you need to tell Octave which strategic context to use for messaging. You have two main approaches.

Approach 1: Using Playbooks (Recommended for Most Users)

Playbooks are tight, pre-defined strategies you've already built in your Library. Each playbook bundles together specific personas, use cases, segments, and reference customers into one focused go-to-market narrative.

Why Playbooks are powerful: They're opinionated. You've already made strategic decisions about who to target and how to position to them. Playbooks let you reuse that thinking without reconfiguring every element.

Option A: Best Match Playbook Mode (Auto)

When the agent runs on your list, it will review the prospect's LinkedIn profile and company domain, analyze which playbook from your Library is the best fit, and use that focused strategy to guide the messaging. Best for when you have multiple tight playbooks defined and want Octave to intelligently pick the right strategy for each prospect.

Example: You have three playbooks, "Series B Tech Companies," "Enterprise Healthcare," and "Mid-Market SaaS." With Best Match mode, Octave looks at each prospect and automatically chooses which playbook to use. No two prospects get the same messaging, it adapts based on their fit with each playbook.

Option B: Best Of Playbook Mode (Auto)

Same as Best Match but you limit Octave to choose from a subset of playbooks you provide.

Option C: Single Playbook Mode (Manual)

You hard-code the agent to use one specific playbook for your entire list. Best for when your list is already highly focused (for example, all VPs of Engineering at Series B companies or all Marketing Directors in Healthcare).

How to set it up: Switch from Auto mode to Manual mode, select the specific playbook for your focused audience, and every prospect on your list gets messaging tailored to that playbook.

Approach 2: Customizing Context Manually (Advanced)

If you don't want to use pre-built playbooks, you can manually customize exactly which context elements to use. This gives you granular control but requires more setup.

Instead of using a playbook, you're bundling together specific elements from your Library: Personas, Use Cases, Segments, Reference Customers, and Proof Points.

For each element, you can choose between:

  • Auto Selection — Let Octave decide which element is best for each prospect. Example: "Choose the best persona based on job title and responsibilities." Octave analyzes each prospect and picks the most relevant persona automatically.
  • Manual Selection — Specify exactly which elements to use. Example: "Only create messaging for the VP of Engineering persona." or "Only use these three use cases: API Integration, Real-Time Analytics, Compliance." Everyone on your list gets messaging built around those exact elements.

You can also mix and match — set some elements to Auto (Octave picks) and others to Manual (you specify).

Value Props Toggle

Value Props ON: Octave generates persona-specific versions of your story for each contact. This creates personalized messaging: each persona sees different value propositions.

Value Props OFF: Everyone in your list receives the same value proposition and story. This creates more consistent messaging but less personalization.

Best practice: Toggle ON for diverse lists. Toggle OFF for highly homogeneous lists.

Which Approach Should You Use?

Use Best Match / Best Of Playbook Mode (Auto) if: You have tight playbooks you trust, you want to save configuration time, your list has diverse personas or use cases, you want Octave to intelligently route messaging.

Use Single Playbook Mode (Manual) if: Your list is already very focused, everyone fits into one clear strategy, you want consistency across the entire campaign, you know exactly who you're targeting.

Use Customized Context if: You need granular control over context elements, your use case doesn't fit a standard playbook, you want to experiment with different context combinations, you're comfortable with extra configuration.

Pro Tip: If you have strong playbooks defined, use Best Match Mode, it saves time and produces excellent results. Only go custom if you have a specific reason to.

Writing Style
Tone, structure, CTAs, methodology, and training examples

After selecting your context, the next step is configuring the writing style. This is your canvas to be as opinionated as you want with specific copywriting nuances. The writing style determines exactly how your emails will sound, look, and flow.

Best Practice: Start with Defaults

Before tweaking anything, run the agent with the default writing style first. It gives you a baseline to work from. Hit "Run Test" and see what Octave produces out of the box with standard configurations. This way, you can see what the agent produces with your context and offering, make deliberate informed changes, and avoid guessing, you'll know exactly what needs adjustment.

Global Email Settings

Tone

  • Casual — Conversational and friendly. Good for younger companies, creative industries, or founder-to-founder outreach.
  • Neutral — Professional but approachable. Most common; works across most industries and personas.
  • Formal — Traditional business tone. Best for enterprise, legal, or highly regulated industries.

Reading Level: Controls vocabulary complexity. Recommended: High School. This level keeps your emails professional yet clear, smart people should understand you easily without confusion or jargon.

Email-Level Configurations

Here's the powerful part: each email in your sequence can have completely different configurations. Email 1 might be casual and short. Email 2 might be neutral and medium. Email 3 might be formal and long. You have complete flexibility.

Subject Line Settings

Configure how your subject lines are generated: include the person's name, include the company name, make it all lowercase (often improves open rates), add custom instructions for formatting.

Custom instruction examples: "Always use this format: [Company Name] + [Benefit]." or "Start with a question mark and number the sequence (Re: 1, Re: 2, etc.)."

Email Body Settings

Basic Controls: Length (Short/Medium/Long), Sentences (e.g., "7-9 sentences"), Paragraphs (e.g., "3-4 paragraphs").

Methodology (How Octave Structures the Email)

This controls how Octave writes your emails. Choose from three options:

  • Octave Methodology — Octave's opinionated structure that follows proven best practices. Starts with the problem, avoids clichés like "Hope this finds you well," and follows a clear flow. Start here by default. It's battle-tested and high-performing.
  • Mixed Methodology — Uses Octave's opinionated structure but lets you add specific rules. Use when you mostly like Octave's approach but want extra control. Examples: "Don't use these jargon terms", "Avoid starting sentences with 'Additionally'", "Keep paragraphs to two sentences max."
  • Custom Instructions — Gives heavier weight to your own structure. Use when you have strong opinions about format. Examples: "First line should mention [X]", "Second line should say [Y]", "Always include [Z] in paragraph three."

Focus

Controls what the email emphasizes:

  • Pain — Highlights the problem. Great for early emails.
  • Benefits — Emphasizes outcomes. Great for later-stage emails.
  • Neutral — Balanced mix of both. Most common.

Call to Action Settings

Every email needs an ending. Octave offers preset CTAs or allows custom ones.

Preset Options: Initiate conversation, Find time to connect, Request introduction, Ask for feedback, Politely disengage.

Best Practice CTA Strategy based on top-performing campaigns:

  • Emails 1-2: Don't ask for time. Use "Initiate conversation" or a soft lead magnet ask.
  • Emails 3-4: Ask for time, "Find time to connect" performs well here.
  • Final email: Politely disengage. Giving an "out" builds trust and drives replies.

Custom CTAs: Use these when you want something not in presets. Example: "Hey, can I send you a guide on how a customer similar to you solved this problem?" This is a soft ask that's easy for prospects to say yes or no to. Select "Custom" instead of a preset, write clear instructions for structure, and describe the lead magnet or ask you want to use.

Training Examples (Most Important)

Training examples are the most powerful part of writing style configuration.

They're "golden examples": real emails you'd send to one person. Octave uses that one-to-one example to scale your tone, logic, and flow across your entire list. You're not creating a template. You're showing Octave how you think.

How to add Training Examples: Click "+" under Training Examples, add your golden example for that email, repeat for each email (optional).

What makes a great training example:

  • Highly relevant to a specific person (not generic)
  • Matches the tone and style you want replicated
  • Demonstrates how you'd reference persona, use case, or company details
  • Shows your preferred structure and flow

Example: Instead of "Hey, we help companies with X", try: "I noticed you just hired a VP of Engineering. That usually means you're scaling the team and running into data pipeline issues. We worked with [similar company] on this and helped cut data latency in half while saving 40 hours a week. Worth a quick conversation?"

This tells Octave: use research signals, mention similar companies, lead with outcomes, and end with a soft ask.

Key insight: You're teaching Octave your approach. It will intelligently adapt it to everyone else on your list.

Configuring Individual Emails

Each configuration option can differ for every email in your sequence. Example strategy:

  • Email 1: Casual tone, short, Octave Methodology, Pain focus, "Initiate conversation" CTA
  • Email 2: Neutral tone, medium, Mixed Methodology, Neutral focus, Lead magnet CTA
  • Email 3: Neutral tone, medium, Octave Methodology, Benefits focus, "Find time to connect" CTA
  • Email 4: Formal tone, short, Custom Instructions, Benefits focus, "Politely disengage" CTA

You have full flexibility to design your rhythm.

Pro Tip: Don't try to perfect everything before the first test. Run it with defaults, see what happens, and then make targeted improvements. Iterate, don't over-engineer.

Testing & Deployment
Test, iterate with chat, save presets, and deploy at scale

Now that you've configured your agent, it's time to test it, tune it up, and launch it in your automation tool. This section covers how to iterate on your outputs and integrate Octave into your workflow.

Step 1: Run Your First Test

Hit "Run Test" to see what your agent produces with the current configuration.

What to look for:

  • Does the tone match your brand?
  • Is the structure right?
  • Are the CTAs appropriate?
  • Does it use context effectively?
  • Does it feel like something you'd actually send?

Don't expect perfection on the first run. This is your starting point for iteration. Even if you like 70% of what you see, that's a great baseline to build from.

Step 2: Improve Outputs with Chat

Instead of manually adjusting configurations, use the Improve Email feature to chat directly with the agent.

How to use Improve Email:

  1. Click the "Improve Email" button on any output
  2. Describe what you want to change (e.g., "Split up the paragraphs for better readability" or "Make the second paragraph more specific to their company")
  3. Select which emails to improve (or apply to all)
  4. Review the configuration changes Octave makes
  5. Choose to either Approve Changes (save updates) or Approve and Rerun (save and test again immediately)

Key benefit: You're chatting with the agent to improve outputs instead of manually tweaking settings. Octave applies your feedback automatically to the writing style, saving time and keeping iterations fast.

Step 3: Turn Good Outputs into Training Examples

When you see an output that's about 90% there and you like it but want to tweak it, convert it into a Training Example.

How to add outputs as Training Examples:

  1. Find an output you mostly like
  2. Click "Add to Training Examples"
  3. The email appears in your Training Examples section
  4. Edit it to make it perfect
  5. Save it as your golden example

Why this matters: Training examples are the most powerful way to teach Octave your exact voice. Starting with a strong output and refining it is faster than writing from scratch. This creates a feedback loop: Octave generates → you refine → it becomes a training example → future outputs improve.

Step 4: Save Your Writing Style as a Preset

Once you have a writing style you love, save it to reuse across future agents.

How to save a preset: Go to the top of the Writing Style panel, click "Save Style", name your preset (e.g., "Cold Outbound – Tech Buyers" or "Warm Inbound – Enterprise"), and load it in future agents as needed.

Benefit: You shorten the path to high-quality copy. Instead of configuring from scratch each time, load your preset and you're 80-90% there instantly.

Using Saved Presets: When building a new agent, go to the Writing Style step, load your saved preset, the agent immediately produces copy that matches your tone, and make minor tweaks for the specific campaign if needed.

Step 5: Test in Your Automation Tool

Before launching at scale, test your agent on a small sample.

Best practice: Run on 5-10 rows first to check how it performs across different prospects, catch issues before full rollout, and make final writing style adjustments.

If you're using Clay: Add the Octave enrichment column to your table, select 5-10 rows to run, review outputs across personas and companies, and confirm the agent adapts correctly to each context.

Step 6: Deploy to Your Workflow

Once outputs look good, integrate Octave into your automation workflow.

Example: Clay Integration:

Your data table should include: First name, Job title, Company domain, LinkedIn profile, Company name.

Setup steps:

  1. Add the Octave column or node to your Clay table
  2. Run on 5-10 rows to verify results
  3. Map the outputs to your campaign tool: Email campaigns (Instantly, Smartlead), LinkedIn campaigns (HeyReach), Both (SuperSend, LaGrowthMachine, Lemlist)

Pro Tip: Extract Email Fields: Open the Octave payload, extract emails and subject lines as individual columns, and map those directly into your campaign tool. This keeps your view clean and ensures accurate delivery.

Step 7: Deploy at Scale

When you're confident in the results: remove the row filter, run the agent on your full list, generate emails for every prospect, map the outputs to your campaign tool, and launch your sequence.

In Short: The Sequence Agent is designed for iteration. Run it, see what comes out, refine, and repeat until it's production-ready. The combination of chat-based refinements, training examples, testing with profiles in Octave, and testing on profiles in Clay should help you tailor your agent to match your specific and desired style.

Credit Usage: Pulse = 1 credit per email. Echo = 2 credits per email. Harmony = 3 credits per email.

Continue with the Sequence Agent
Run with MCP
Next Agent
Content Agent
Next Section
Connect Octave Everywhere

Sequences via MCP

With Octave connected as an MCP server, you can run a pre-built sequence agent or create one on the fly through natural conversation. Build, test, iterate, all without leaving your MCP client.

Run a Pre-Built Agent
"Run my 'Cold Outbound' sequence agent for Sarah Chen, VP Marketing at Stripe." Claude calls run_email_agent with your saved agent, applying your playbook, persona targeting, per-email instructions, and style preferences.
Create On the Fly
"Write a 3-email cold outreach sequence for VP Marketing personas. Use my 'Series B SaaS' playbook. Tone should be confident but not pushy." Claude calls generate_email with your playbook and persona context, no saved agent needed. Produces the full sequence on the spot.
Build Example Sequences
"Generate 3 example sequences for different prospect types, a VP at a startup, a Director at enterprise, and a founder. I want to see how the agent handles each." Claude runs your agent against each profile so you can compare the output across different scenarios before going live.
Test Against Real Profiles
"Enrich Sarah Chen at Stripe, then run my sequence agent against her profile. Show me what the outreach would look like." Claude chains enrichment into your sequence agent so you can see how it performs with real prospect data, not hypotheticals.
Iterate and Adjust
"The second email feels too salesy. Make it more consultative and re-run for the same prospect. Also shorten email 3." Claude updates your agent's configuration based on your feedback and regenerates, giving you a tight build-test-refine loop without leaving the conversation.
Next Agent
Content Agent
Next Section
Connect Octave Everywhere

What's New for the Sequence Agent

View all updates →
Jan 2026

Sequence Agent Updates

Compressed UI, model selection, native deep research, context fallback logic, and granular content controls.
  • Compressed Agent UI — Cleaner layout showing config, context, and writing style at a glance.
  • Model Selection — Trade off cost and reasoning power per use case.
  • Native Deep Research — Pull in live, external information with guided research directives.
  • Context Fallback Logic — Define behavior when no playbook is a strong fit.
  • Granular Content Controls — Include or skip reference customers and proof points per sequence.
Jan 2026

Bare Metal Mode

Strip away opinionated guardrails for direct control over how the model reasons and writes.
  • Full creative control — Dictate exactly how the model approaches content.
  • Advanced prompt engineering — When custom instructions are detailed enough that defaults get in the way.
  • Experimentation — Test new angles with raw model output before layering in structure.
Dec 2025

Brand Voice

Reusable brand voices with core traits, guiding principles, and tonal qualities across all sequence runs.
  • Structured voice controls — Core traits, guiding principles, tonal qualities, key terms.
  • Agent-level application — Select once, applies across all generated content.
  • Reduced prompt overhead — No more repetitive custom instructions for brand consistency.
Dec 2025

Smart CTAs

Dynamic CTA selection based on recipient profile with a reusable CTA library.
  • Dynamic selection — CTAs matched at runtime based on recipient profile data.
  • CTA Library — Reusable CTAs with usage scenarios and hyperlinks.
  • Auto or manual mode — Let the agent pick or restrict to a specific CTA.
HomeDeploy AgentsRuntime Context
Put It to Work

Runtime Context

Pass dynamic, per-record data into any agent at runtime, from job postings and tech stack to CRM history and engagement signals.

What Runtime Context Is

Runtime context is any line-by-line data you bring into an Agent at runtime. It can be anything you know about your leads, customers, or accounts, whether from external research, enrichment tools, or your own systems.

Hiring Signals
Open job postings, team growth, new VP hires
Company Intel
Office locations, expansion regions, recent funding
Tech & Product
Tech stack, recent product launches, integrations
Engagement Data
Call notes, CRM history, product usage, past campaigns

This data is not fixed in your Playbook or Library. Instead, it's passed in dynamically when the Agent runs, allowing Octave to weave these details into its reasoning or generation process.

How to Use Runtime Context

Runtime context can be added to any Agent in Octave. Each Agent includes a field where you can pass in situational information so Octave can make smarter, more tailored decisions.

You usually combine two parts: plain language context that explains what the data means, and dynamic references that point to specific fields in your source data.

For example, if your lead list includes a column showing logins:

"This person has logged into our app {{last_month_logins}} times in the past month."

The words guide the Agent's interpretation, and the {{last_month_logins}} value updates dynamically for each row. This makes instructions both readable and scalable.

In Sequence Agents

In Sequence Agents, you can also give tactical instructions on how that context should be used in your messaging. These can be applied at the sequence level or on a per-email basis.

"In the first sentence of the email, mention that we noticed they logged in {{last_month_logins}} times this month."

This tells the Agent exactly where and how to use the context while keeping your Playbooks and Library consistent and reusable across campaigns.

Next
Content Agent
Next Section
Connect Octave Everywhere
HomeDeploy AgentsContent Agent
Put It to Work

Content Agent

Your adaptable content engine. Generate anything that needs to sound like you: LinkedIn DMs, website copy, blog posts, sales decks, enablement materials, all grounded in your Library.

Overview
Deep Dive
Run with MCP
What's New

Think of the Content Agent as a freeform agent that has access to your entire Library and can be configured for any repeatable content job. You define the task, the format, and the context it should pull from, then run it once or hundreds of times across different inputs.

How It Works

The Content Agent combines your Library context (products, personas, proof points, competitive positioning) with your custom instructions to produce content. Because it already has your company knowledge loaded, you skip the usual cycle of prompt engineering and re-explaining who you are with every generation.

You focus on what you want to create. The agent handles grounding it in your actual messaging.

What You Can Do

Outbound Messages
LinkedIn DMs, one-off emails, intro notes: personalized to specific prospects and pain points
Marketing Copy
Landing pages, product descriptions, ad copy, campaign messaging across segments
Blog Posts & Articles
Long-form content grounded in your positioning, proof points, and market perspective
Enablement Materials
Battle cards, objection handlers, cheat sheets, one-pagers: kept current as your Library evolves

Content Agent vs. Sequence Agent

The Sequence Agent is purpose-built for multi-email outbound sequences, coordinated cadences designed for cold outreach and follow-up flows.

The Content Agent is open-ended. It generates any single piece of content in any format. It's not constrained to email sequences or outbound workflows, you give it a job and a format, it produces the output.

Test in Octave, deploy at scale. Refine outputs in the playground, then run the same agent programmatically through Clay, n8n, Cargo, or Zapier.

Continue with the Content Agent
Deep Dive
Next Agent
Call Prep Agent
Next Section
Connect Octave Everywhere

Generate Contextual Content

The Content Agent takes your natural-language instruction and your Library context to produce complete, on-message copy for any surface: outbound, marketing, internal documentation, or creative exploration.

  • Outputs are grounded in your products, personas, proof points, competitive positioning, and tone from day one
  • Understands the nuance between different personas and segments, messaging for a VP of Sales sounds different from messaging for a CMO, even when describing the same product capability

Refine and Control Outputs

Iterate quickly on tone, length, and structure using short natural edits like "make it more conversational," "summarize this in one paragraph," or "focus more on ROI."

  • Output format — text, JSON, HTML, and more depending on what you need
  • Model power — choose the right level of AI horsepower for your content
  • Core context — select which playbooks, personas, proof points, use cases, and competitive intelligence to feed in
  • Manual content selection — fine-tune which specific Library elements are used, ensuring the agent draws from the right positioning for your target
  • Web search — toggle on to have the agent research and incorporate real-time context from the web
  • Brand voice — connect your brand voice (coming soon) to ensure consistency across all generated content at scale

The Bigger Picture

Scaling content creation without sacrificing quality has historically required either hiring more writers or accepting generic output. The Content Agent offers a third path: use your institutional knowledge to create relevant, on-brand content at the speed and volume you need.

Whether you're writing one thing or deploying hundreds of variations across different channels and segments, everything stays rooted in your actual company knowledge.

Continue with the Content Agent
Run with MCP
Next Agent
Call Prep Agent
Next Section
Connect Octave Everywhere

Build & Test with MCP

With Octave connected as an MCP server, you can run your saved content agents, build new ones, and refine them, all through conversation.

Run a Saved Agent
"Use my 'Competitive Displacement' content agent to write a follow-up email for the VP of Eng at Acme Corp. Focus on migration ease." Claude runs your existing saved agent with the prospect details and context you provide. Same agent, same configuration, just a new input.
Set Up a New Agent
"Create a content agent for writing LinkedIn posts. Use my 'Product Launch' playbook and target marketing leaders. Tone should be insightful but approachable." Claude creates the agent with your playbook, persona, and style preferences, no need to navigate through setup screens.
Generate Test Content
"Generate 3 examples — a LinkedIn post, a follow-up email, and a short case study intro. I want to see how the agent handles different formats." Claude runs your agent across different content types so you can evaluate the output before deploying at scale.
Test with Real Context
"Enrich Acme Corp first, then write a personalized case study pitch email using what you learn about their tech stack and growth stage." Claude chains enrichment into your content agent so you can see how it performs with real company data, not hypotheticals.
Iterate and Adjust
"The tone is too formal. Make it more conversational and add a stronger CTA. Also switch to the 'Competitive Displacement' playbook." Claude updates your agent's configuration based on your feedback and regenerates, a tight build-test-refine loop without leaving the conversation.
Next Agent
Call Prep Agent
Next Section
Connect Octave Everywhere

What's New for the Content Agent

View all updates →
Jan 2026

Bare Metal Mode

Strip away opinionated guardrails for direct control over how the model reasons and writes.
  • Full creative control — Dictate exactly how the model approaches content.
  • Advanced prompt engineering — When custom instructions are detailed enough that defaults get in the way.
  • Experimentation — Test new angles with raw model output before layering in structure.
Dec 2025

Brand Voice

Reusable brand voices with core traits, guiding principles, and tonal qualities across all content runs.
  • Structured voice controls — Core traits, guiding principles, tonal qualities, key terms.
  • Agent-level application — Select once, applies across all generated content.
  • Reduced prompt overhead — No more repetitive custom instructions for brand consistency.
HomeDeploy AgentsCall Prep Agent
Put It to Work

Call Prep Agent

Generate tailored briefing docs and discovery scripts for specific accounts and stakeholders. Out-of-the-box structure with per-section custom instructions.

Overview
Deep Dive
Run with MCP

The Call Prep Agent gives you a structured, meeting-ready briefing document out of the box: discovery questions, call script, objection handling, person brief, company brief, and relevant case studies. Everything is grounded in your Library and enrichment data.

Each output section supports Custom Instructions — natural-language guidance that shapes what that section produces (e.g. tell the discovery questions section to always ask about contract timelines, or tell the call script to lead with an industry insight instead of a product pitch).

If the standard sections cover 80% of what you need, the Custom Task tool is your escape hatch: a separate tool that goes and does an entirely new job alongside the built-in outputs. Competitive deep-dive, pricing analysis, regulatory checklist, whatever your team's workflow demands.

Continue with the Call Prep Agent
Deep Dive
Next Section
Connect Octave Everywhere

How the Agent Works

The Call Prep Agent takes your target input (a person, company, or both), pulls context from your Library and any enabled tools, then generates each output section independently: each section gets the full context window focused on its specific job.

Input (Person / Company)
Library + Tool Context
6 Output Sections

The Six Output Sections

Every call prep document includes these sections, each generated from your Library context and enrichment data. Toggle any section on or off.

Discovery Questions
Tailored questions based on the prospect's company, role, and the playbook you're running: not generic templates
Call Script
Structured talking points with opening, value prop delivery, and transition prompts tuned to the specific meeting
Objection Handling
Anticipated objections with recommended responses, grounded in your competitive positioning and proof points
Person Brief
Background on the specific stakeholder: role context, likely priorities, and relevant talking points
Company Brief
Company overview, recent news, tech stack, growth signals, and strategic context for the conversation
Relevant Case Studies
Matched proof points and customer stories from your Library that fit this prospect's situation

Custom Instructions — Shaping Each Output

Each output section has a + Custom Instructions toggle. These instructions shape what that specific section produces: the tone, the framing, the specific things you always want included. They persist on the agent, so you set them once and every future run inherits your preferences.

Examples of effective custom instructions:

Discovery Questions
"Always include 2 questions about their current tech stack and 1 about budget timeline. Frame questions using the Sandler methodology."
Call Script
"Open with a relevant industry insight, not a product pitch. Keep the intro under 30 seconds. Include a clear next-step ask."
Objection Handling
"Focus on the top 3 objections for this persona. Always include a proof point or case study reference in the response."
Company Brief
"Prioritize recent funding, leadership changes, and any competitive tools in their stack. Include a section on likely pain points."

Tools

The Call Prep Agent can pull from multiple data sources to enrich the briefing. Toggle tools on or off based on what context you want.

Deep Research
Web research for enhanced context: recent news, funding rounds, org changes, and public information
CRM Activity
Pull contact and company data from your connected CRM for deal history, past interactions, and account context
Custom Task
A separate tool that goes and does an entirely new job: competitive analysis, pricing comparison, regulatory checklist, anything the standard sections don't cover

Custom Task vs. Custom Instructions

These are different things that work together:

  • Custom Instructions shape the output of existing sections, tell the discovery questions to focus on budget, tell the call script to use a specific methodology. The section still does its original job, just tuned to your preferences.
  • Custom Task is a separate tool that goes and does a completely different job alongside the standard outputs. It runs its own analysis and returns structured JSON. Use it when you need something the six built-in sections don't cover at all.

Think of it as: Custom Instructions make the 80% better. Custom Task handles the other 20%.

Agent Configuration

When setting up your Call Prep Agent, you configure:

Methodology
Octave: the agent uses your Library as its knowledge foundation
Model
Pulse (default): balanced speed and quality for briefing generation
Tools
Toggle Deep Research, CRM Activity, and Custom Task on or off depending on what context you want pulled in
Outputs
Enable or disable each of the 6 sections. Add custom instructions to any enabled section.

Setting Up the Agent

  1. Create the agent — Navigate to Agents, select "Build New Agent", and choose Call Prep Agent. The six output sections are enabled by default.
  2. Enable your tools — Toggle on Deep Research for web context, CRM Activity if you have a CRM connected, and Custom Task if you need an additional analysis job.
  3. Add custom instructions — Click "+ Custom Instructions" on any output section. Write natural-language guidance for how that section should be shaped for your sales process.
  4. Test with a real prospect — Search for someone you're actually meeting with soon. Review each section and iterate on your custom instructions until the output matches what you'd want to walk into a call with.

Use the Output

Call prep documents can feed directly into your team's workflow:

  • Share the full briefing with your AE before a demo or discovery call
  • Extract discovery questions into your meeting notes template
  • Use the objection handling section as pre-call coaching material
  • Run call prep at scale through Clay or n8n to brief your entire team before a pipeline blitz
  • Feed the company brief into your CRM as pre-call context for any rep touching the account

Credit Usage: Credit cost depends on which tools you enable. Deep Research costs 8 credits, CRM Activity costs 10 credits, and Custom Task costs 5 credits. Base call prep generation is included.

Continue with the Call Prep Agent
Run with MCP
Next Section
Connect Octave Everywhere

Call Prep via MCP

With Octave connected as an MCP server, you can run call prep through natural conversation. Two modes: run a pre-built agent with your saved configuration, or generate one on the fly by describing who you're meeting. If you have other MCPs connected (CRM, Slack, etc.), you can push the outputs directly into those systems.

Run a Pre-Built Agent
"Run my 'Discovery Call' agent for the CTO at Figma." Claude calls run_call_prep_agent with your saved agent, applying your custom instructions, tool preferences, and output configuration.
Generate On the Fly
"Prep me for a call with Sarah Chen, VP Marketing at Lattice. We're running the Enterprise playbook." Claude calls generate_call_prep with the person and company context, no saved agent needed. Produces a full briefing document with all six sections.
Prep with Extra Context
"Prep for my call with the Stripe team. They just announced a new payments API and their contract with our competitor is up for renewal in Q2." Claude passes your additional context into the briefing, deal-specific intelligence you already have gets woven into every section.
Batch Prep
"I have 5 discovery calls this week. Prep all of them: Sarah at Lattice, Mike at Figma, Diana at RevPartners, Tom at Stripe, and Lisa at Notion." Claude runs call prep for each meeting sequentially, giving you a complete briefing packet for the week.
Push to Other MCPs
"Prep for my Lattice call, then post the company brief to the #lattice Slack channel and log the discovery questions in HubSpot as a note." Claude generates the call prep, then uses your connected Slack and CRM MCPs to push outputs where your team will actually see them.
Next Section
Connect Octave Everywhere
HomeConnect Octave Everywhere
Put It to Work

Connect Octave Everywhere

MCP gives every tool on your desk direct access to your GTM brain. Build your library from your IDE, research prospects mid-workflow, generate content without leaving your editor.

What is MCP?

The Model Context Protocol (MCP) lets AI-powered tools connect directly to Octave. Instead of copying data between tabs, your tools talk to Octave natively — full read/write access to your Library, Playbooks, and Agents, all through conversation.

Your Tool (Claude, Cursor, etc.)
MCP
Octave (Brain)

Explore

Setup

Get Connected

Step-by-step setup guides for Claude Desktop, Claude Code, Cursor, Windsurf, and n8n. Pick your client and connect in minutes.

Claude Desktop
Claude Code
Cursor
Windsurf
Use Cases

What You Can Do

Build your library conversationally, run agents from your IDE, research prospects mid-workflow, generate content without leaving your editor.

Plugin

Skills Library

The LFGTM plugin adds 30+ guided GTM skills on top of MCP for Claude Code. Battlecards, audits, research, content — all orchestrated.

Next Section
Connect Your Stack
HomeConnect Octave EverywhereWhat You Can Do
Put It to Work

What You Can Do with MCP

Once connected, Octave becomes a native part of your workflow. Here's what opens up.

Build Your Library Conversationally

Create and refine every library element through natural conversation. Add personas, offerings, use cases, segments, and competitors by describing them — no forms, no clicks. Paste in call transcripts, articles, or competitor pages and ask Octave to extract and structure the intelligence.

Example
"Create a persona for VP of Revenue Operations focused on pipeline velocity and forecast accuracy." Claude calls create_entity with the structured persona. Works for all 7 entity types.

Run Agents from Your IDE

Trigger any Octave agent — enrichment, qualification, sequence, content, call prep — directly from Claude Code, Cursor, or any MCP client. No browser tab, no context switching. Results come back inline.

Example
"Qualify jane@acme.com against my Enterprise ICP and generate a 3-step outbound sequence." Claude runs the qualification agent, then the sequence agent, all from your terminal.

Research Prospects Mid-Workflow

Pull enrichment data, company context, and qualification scores without leaving your current task. Ask about a prospect's fit, get a company brief, or check what use cases fit, all inline.

Generate Content Without Leaving Your Editor

Need a follow-up email? A LinkedIn message? A battlecard? Ask for it in conversation and Octave generates it from your Library context — same quality as the web UI, zero tab-switching.

Next
Skills Library
HomeConnect Octave EverywhereSkills Library
Put It to Work

Skills Library

The LFGTM plugin adds 30+ guided GTM skills on top of the raw MCP connection. MCP is the engine — the plugin is the dashboard.

What the Plugin Adds

While MCP gives you raw tool access (create entities, run agents, etc.), the plugin orchestrates those tools into complete, guided workflows. Install via GitHub for the full conversational GTM toolkit.

Featured Skills

/octave:library
Guided library building with best practices baked in
/octave:audit
Health checks for gaps, stale content, and duplicates
/octave:generate
Content generation with mode selection (agent, API, or direct)
/octave:research
Deep account research and call prep
/octave:battlecard
Competitive intelligence and displacement campaigns
/octave:messaging
Positioning, narratives, and messaging matrices
/octave:campaign
Multi-channel campaign planning and content generation
/octave:pipeline
Deal coaching with diagnosis and next-step recommendations
/octave:prospector
Find, enrich, and qualify prospects against your ICP
/octave:deck
Presentation builder powered by your Library

And 20+ more specialized skills for enablement, training, ABM, proposals, microsites, win/loss analysis, and more.

Next Section
Connect Your Stack
HomeConnect Your Stack
Put It to Work

Connect Your Stack

Octave connects to your orchestrators and execution tools. The brain powers your entire GTM stack from one living context.

The Integration Layers

Octave sits between your strategy and your execution. Connect at two levels: orchestrators that move data and trigger workflows, and execution tools that send the emails and run the campaigns.

Octave (Brain)
Orchestrators
Execution Tools

Choose a Connection

Data & Workflow Layer

Orchestrators

Move data, enrich records, trigger Octave agents at scale. The plumbing that connects your brain to the world.

Clay
n8n
Cargo
Zapier
Email & Campaign Layer

Execution Tools

Send the emails, run the sequences. Map Octave-generated content into your sending platforms via orchestrators.

Instantly
Smartlead
HeyReach
SuperSend
Video: Connecting Your Stack
Next Section
Make It Smarter
HomeConnect Your StackOrchestrators
Put It to Work

Connect to Orchestrators

Orchestrators move data, trigger workflows, and run Octave agents at scale. They're the plumbing between the brain and your execution tools.

Clay
Native integration — connect your API key, select agents, map fields, auto-update

Once you've created and configured your Agents in Octave, the next step is activating at scale by connecting to your lead list in Clay.

Setup Steps

  1. Add Octave enrichment — In Clay, go to Add Column > Add Enrichment > search "Octave".
  2. Connect your account — Click Add Account, name it, and paste your Octave API key (found via the "Use in Clay" button in any Octave Agent).
  3. Test and save — Click Test Account, then Save.
  4. Select your agent — Open the Agent/Experiment dropdown under Setup Inputs. Your Octave agents should appear in the list.
  5. Map your fields — Connect Clay columns to agent inputs (name, title, company, LinkedIn, etc.).
  6. Enable auto-update — New leads automatically flow through Octave and receive generated content.

You only need to wire the account once. All future Octave enrichments in Clay will use this connection.

Clay Lite
Use Octave inside Clay on Clay credits — no separate account needed

Octave Lite runs inside Clay's native integration on Clay credits. It generates personalized, multi-step email sequences without needing a separate Octave account. Great for quick, contextual messaging directly from Clay.

Setup Steps

  1. Prepare your list — Start with a Clay table containing standard contact and company fields.
  2. Add the enrichment — Add Column > Add Enrichment > search "Octave" > select the Lite version (identified by "Clay credits" notation).
  3. Configure sequence settings — Email count (1–7), length (short/medium/long), type (cold/warm/inbound/custom), primary goal, and optional unique subject lines.
  4. Provide brand examples — Paste a great first email and final email that reflect your brand voice. The AI personalizes while maintaining your tone.
  5. Add instructions — Brand voice guidance, preferred terminology, key angles, compliance notes.
  6. Connect product context — Website URL, product name, value proposition, differentiators, social proof, and supporting materials.
  7. Map contact fields — LinkedIn profile, first name, job title, company domain, and company name.
  8. Add personalization data — Link columns with role openings, funding info, tech stack, or relevant news for deeper customization.
  9. Run and review — Save, run the enrichment, and review generated emails and subjects.
n8n
Connect via HTTP nodes — credentials, agent selection, lead mapping, runtime context

Connect Octave to n8n and run a Sequence Agent as part of an outbound workflow. The integration assumes you already have lead data flowing into n8n: the focus is on triggering Octave and getting personalized sequence outputs back.

Setup Steps

  1. Configure Octave credentials — Set up your Octave API credentials within n8n.
  2. Choose your agent — Select which Octave agent to run in the workflow.
  3. Map lead data — Input lead information such as name, title, and LinkedIn profile.
  4. Add runtime context — Optionally include extra context for enhanced personalization.
  5. Route outputs — Pass Octave-generated content to your email platform or next workflow step.
Cargo
API-based integration for orchestrating Octave agents within Cargo workflows

Connect Octave to Cargo via API key and map fields to orchestrate agents within Cargo workflows. Useful for teams already running their data enrichment and routing through Cargo.

Video coming soon

Zapier
Trigger Octave agents from any Zapier-supported event

Trigger Octave agents from any Zapier-supported event. Useful for lightweight automations and connecting Octave to 5000+ apps. Route CRM updates, form submissions, or any webhook into Octave for real-time content generation.

Video coming soon

Next
Execution Tools
HomeConnect Your StackExecution Tools
Put It to Work

Connect to Execution Tools

Execution tools send the emails and run the campaigns. Octave-generated content flows through your orchestrator (usually Clay) into these platforms.

The pattern is always the same: Octave generates the content (via orchestrator) > content maps to custom variables in your execution tool > execution tool sends at scale. The brain writes, the tools deliver.

Instantly via Clay
Map Octave outputs to Instantly custom variables for automated email delivery

This integration routes Octave-generated email content through Clay into Instantly campaigns, eliminating manual copy-pasting. Subject lines and email bodies map to Instantly custom variables for personalized, multi-step outreach at scale.

Setup Steps

  1. Run Octave in Clay — Execute your Octave agent to generate cold emails and subject lines. Optionally separate into distinct columns (SubjectLine1, Email1, etc.).
  2. Connect Instantly to Clay — Select your Instantly account (or add via API Key) and choose your target campaign.
  3. Map basic fields — Complete required fields: Email, First Name, Last Name, Company Name.
  4. Create custom variables — For each email step, create two custom fields — one for subject lines, one for email bodies. Map accordingly.
  5. Test the integration — Disable auto-update, run a test contact, verify custom fields loaded in Instantly's Sequences > Variables section.
  6. Populate sequence with variables — Edit each campaign step in Instantly using placeholders like {{Subject1}} and {{Email1}}.
  7. Preview and launch — Use Instantly's Preview tool to verify rendering, then enable auto-update in Clay.
Smartlead via Clay
Enroll leads into Smartlead campaigns with Octave-generated content via Clay

Export Octave-generated email sequences through Clay and enroll leads into Smartlead campaigns using the "Add Lead to Campaign" enrichment.

Setup Steps

  1. Prepare your Clay table — Ensure it contains lead info (name, title, company, work email) and Octave-generated content (subject lines, body text).
  2. Add the Smartlead enrichment — Add Column > Add Enrichment > select "Smartlead: Add Lead to Campaign."
  3. Configure API key — Retrieve from Smartlead via Settings > API Key, paste into Clay.
  4. Select your campaign — Choose the Campaign ID where leads should be enrolled (must be active in Smartlead).
  5. Map lead data — Connect Clay columns: Email Address, First Name/Last Name, Phone Number, Profile URL.
  6. Map custom variables — Assign Octave outputs to Smartlead variables (e.g., custom1 = Email 1 Subject, custom2 = Email 1 Body).
  7. Test first — Disable auto-update, run a sample, review records for accuracy, then re-enable and run at scale.
HeyReach
LinkedIn outreach automation with Octave-generated messaging

Route Octave-generated messaging into HeyReach sequences for personalized LinkedIn outreach at scale. Connect via API or orchestrator.

Video coming soon

SuperSend
Multi-channel outreach with Octave-generated email and LinkedIn messaging

Connect Octave outputs to SuperSend for personalized email and LinkedIn messaging at scale. Multi-channel outreach powered by your Octave brain.

Video coming soon

Next Section
Make It Smarter
HomeConnect Octave EverywhereGet Connected
Put It to Work

MCP Connections

Octave exposes an MCP server you can connect from any MCP-compatible client. Full read/write access to your Library, Playbooks, and Agents, all through conversation.

Claude Desktop — Custom Connector
One-click connection from the Octave dashboard

Copy the connection link from your Octave dashboard via "Connect Via MCP" and paste it into Claude Desktop's Connectors settings. The fastest way to get connected.

Claude Desktop — Remote MCP
Manual configuration via claude_desktop_config.json

Manually add the Octave MCP server config to your claude_desktop_config.json. Use this method if you prefer direct config file control or need to customize the connection.

Claude Code
Connect via CLI and authenticate with /mcp

Run claude mcp add octave --transport http [your-url] to add the Octave MCP server. Then use /mcp in a new session to authenticate. Full read/write access to your Library, Playbooks, and Agents from the terminal.

LFGTM Plugin (Claude Code)
30+ specialized GTM skills on top of MCP

The LFGTM plugin is an optional layer on top of the Octave MCP connection for Claude Code users. While MCP gives you raw tool access (create entities, run agents, etc.), the plugin adds 30+ guided skills that orchestrate those tools into complete workflows.

Think of it this way: MCP is the engine, the plugin is the dashboard. You don't need the plugin to use Octave via MCP, but it makes complex operations dramatically easier.

What the Plugin Adds

  • /octave:library — Guided library building with best practices baked in
  • /octave:audit — Health checks for gaps, stale content, duplicates
  • /octave:messaging — Framework generation (positioning, narratives, matrices)
  • /octave:generate — Content generation with mode selection (agent, API, or direct)
  • /octave:research — Deep account research and call prep
  • /octave:battlecard — Competitive intelligence and displacement campaigns
  • And 25+ more specialized skills

Install via GitHub for the full conversational GTM toolkit.

Cursor, Windsurf, n8n & Other MCP Clients
Generic MCP setup instructions on octavehq.com

Octave's MCP server works with any MCP-compatible client: Cursor, Windsurf, n8n, and more. The same server URL and authentication flow applies across all clients.

For setup instructions specific to your client, visit your homepage:

Go to your Octave homepage for instructions →
Next Section
Connect Your Stack
HomeMake It Smarter
Close the Loop

Make It Smarter

Octave listens to your calls, emails, and CRM activity. It tags everything against your Library, surfaces what's happening in the field, and suggests updates. The brain gets smarter over time.

Listen & Learn

Activity Feed

Octave listens to calls, emails, and CRM activity, then tags everything against your Library. See which personas showed up, what use cases were discussed, what objections were raised. Search across all activity, drill into individual interactions, and act on suggested Library updates.

Analyze & Report

GTM Explorer

Aggregated reports that answer the questions your team is asking: what objections are we hearing? Which segments are we talking to? What's changing? Run out-of-the-box reports or define your own question. Set any cadence and Octave analyzes calls, emails, and CRM data from that window.

HomeMake It SmarterActivity Feed
Close the Loop

Activity Feed

Octave listens to your calls, emails, and CRM, tags every interaction against your Library, and surfaces what matters.

What Gets Captured

Depending on what you connect, Octave can listen to calls, emails, and CRM activity. Every interaction gets tagged against your Library: which personas were present, what use cases were discussed, which objections came up, what competitors were mentioned.

Browse and Search

The Activity Feed gives you a searchable, filterable view of all captured interactions. Drill into any individual call or email to see exactly what was surfaced: the personas identified, the use cases discussed, the objections raised. Search across everything to find patterns or specific conversations.

Suggested Library Updates

As Octave processes your activity, it identifies gaps and opportunities in your Library. Maybe prospects keep mentioning a use case you haven't defined, or a competitor keeps coming up that isn't in your Library yet. Octave surfaces these as suggested updates, review and accept to keep your brain current with what's actually happening in the field.

Next
GTM Explorer
HomeMake It SmarterGTM Explorer
Close the Loop

GTM Explorer

Aggregated reports that answer the questions your team is asking: run them on any cadence across calls, emails, and CRM data.

Out-of-the-Box Reports

Octave ships with ready-made reports for the questions GTM teams ask most. Each report analyzes your calls, emails, and CRM data from a configurable time window.

Objection Trends
What objections are we hearing? How are they changing over time?
Segment Activity
Which segments are we talking to? Where is activity concentrated?
Use Case Coverage
Which use cases are coming up in conversations? Any gaps?
Competitive Mentions
Which competitors are being discussed? In what context?

Custom Questions

If the out-of-the-box reports don't cover what you need, define your own question. Octave will analyze the same data sources (calls, emails, CRM) and produce an aggregated report on whatever you're trying to understand.

Set a Cadence

Reports can run on any schedule you want: weekly, biweekly, monthly, quarterly. Each run looks at the activity from that time window, so you get a rolling view of what's happening in the field without having to manually pull data.

Related
Activity Feed
HomeWhat's New

What's New

Latest updates, features, and improvements. Click the tags to jump to the relevant section.

Apr 2026

Motions

Motions replace manual Playbook creation with a single step that generates a full ICP matrix — tailored narratives at every Persona x Segment intersection. Layer themed playbooks for campaigns, triggers, and competitive displacement.
Motions

What Changed

The "Create Playbooks" section is now "Build Your Strategy." Instead of manually building individual playbooks, you create a Motion by selecting an offering and a motion type (Net New or Upsell). Octave generates the full ICP matrix with a narrative at every Persona x Segment intersection, plus a Default Playbook covering the entire grid.

Key Capabilities

  • ICP Matrix — Full coverage of your addressable market in one creation step. Every intersection gets its own narrative with target overview, operating landscape, talking points, pains, benefits, and stage-based methodology.
  • Motion-Type Awareness — Net New and Upsell generate fundamentally different narratives because the buyer context differs.
  • Motion Playbooks — Layer themed overlays (Thematic, Milestone, Account, Competitive) on subsets of the grid for focused campaigns.
  • Learning Loop — Library changes and engagement data from calls, emails, and CRM refine narratives over time.
  • Stage-Based Methodology — Resonate (awareness), Elevate (urgency), Compel (action) baked into every cell narrative.

What Stays the Same

Your Library is unchanged — offerings, personas, segments, use cases, and proof points all work the same way. The Library feeds directly into Motions. Existing playbooks continue to work with current agents. Agents can be updated to reference Motions at your own pace.

Feb 2026

Octave + Claude MCP

Octave now exposes an MCP server you can connect from Claude Desktop, Claude Code, Cursor, or Windsurf. Full read/write access to your Library, Playbooks, and Agents: conversationally.
Integrations

Connection Methods

Claude Desktop offers two setup options:

  • Custom Connector (Recommended) — Copy a connection link from your Octave dashboard's "Connect Via MCP" section and paste it into Claude Desktop settings under Connectors.
  • Remote MCP Config — For organizations restricting custom connectors, manually add the MCP configuration to Claude Desktop's config file.

Claude Code — Run claude mcp add octave-myWorkspace --transport http [URL], then use the /mcp command in a new session.

Available MCP Tools

  • Library Tools — Entity management, playbook access, value proposition handling
  • Research — Person/company finding, enrichment, qualification
  • Content Generation — Email, content, and call prep generation
  • Agents — Run saved email, content, enrichment, and qualification agents
  • Configuration — Brand voices and writing styles
  • Resources — Global resource management and semantic search
  • Analytics — Finding and event tracking

Claude Desktop — Custom Connector

Claude Desktop — Remote MCP

Claude Code + LFGTM Plugin

Jan 2026

Qualification Agent Updates

High-effort reasoning mode for complex use cases, native deep research beyond LinkedIn, and advanced scoring controls for better downstream workflow alignment.
Agents

Three Key Updates

  • High-effort reasoning mode — An optional capability that gives the agent more reasoning power for complex use cases with layered qualification criteria.
  • Native deep research — The system can search the web to pull in external signals beyond LinkedIn and the company domain, surfacing context that would otherwise require manual research.
  • Advanced scoring controls — More flexibility in how qualification scores are calculated, making it easier to align outputs with downstream workflows and routing logic.
Jan 2026

Sequence Agent Updates

Compressed UI, model selection, native deep research, context fallback logic, and more granular content controls for reference customers and proof points.
Agents

What's New

  • Compressed Agent UI — A cleaner layout that makes it easier to see where you are in the build flow and how agent configuration, context, and writing style map together.
  • Model Selection — Choose between different models, trading off cost and reasoning power depending on the context and complexity being fed into the agent.
  • Native Deep Research — An optional mode that pulls in live, external information beyond LinkedIn and the company domain, with the ability to guide what research it performs.
  • Context Fallback Logic — Advanced settings that let you define how the agent behaves when no playbook is a strong fit — either falling back to your full library or to a specific playbook.
  • More Granular Content Controls — Additional options to include or skip reference customers and proof points, giving you finer control over social proof across a sequence.
Jan 2026

Resource Center

Centralized hub for connecting websites and Google Drive. Smart content ingestion, live re-indexing, and resource-aware Library creation.
LibraryOptimization

This is the first phase of a larger initiative to enable Octave to automatically stay current as your organization evolves, a system that continuously ingests new information, learns from it, and improves over time.

Key Features

  • Centralized Resource Center — A unified hub for connecting Octave to websites and internal files, including static webpages and live Google Drive documents.
  • Smart Content Ingestion — The system selectively pulls in only the pages and documents relevant to building and maintaining your Library, without clutter from SEO pages or unrelated content.
  • Live Re-indexing — Connected websites and documents support ongoing re-indexing, enabling Octave to detect changes and updates in real-time.
  • Resource-aware Library Creation — When building personas, segments, or use cases, Octave can automatically reference connected resources or be directed to specific pages for additional context.
Jan 2026

Bare Metal Mode — Sequence & Content

Get closer to the LLM. Bare Metal mode removes opinionated guardrails from Sequence and Content Agents, giving you direct control over how the model reasons and writes.
Agents

Octave's default agent modes apply structured guardrails (methodology, tone guidance, formatting rules) to keep outputs consistent and on-brand. Bare Metal mode strips those away, giving you a more direct interface with the underlying model.

When to Use It

  • Full creative control — When you want to dictate exactly how the model approaches a piece of content without Octave's opinionated structure.
  • Advanced prompt engineering — When your custom instructions are detailed enough that the default guardrails get in the way.
  • Experimentation — When you're testing new messaging angles and want raw model output before layering in structure.

Your Library context, Playbooks, and runtime context still flow in. Bare Metal just removes the scaffolding around how the model uses them.

Jan 2026

Custom Tasks for Enrich, Qualify & More

Agents like Enrich and Qualify handle a defined set of jobs. Custom Tasks let you bolt on one more, like a mini Content Agent inside any agent.
Agents

Enrich, Qualify, and other structured agents each perform a specific set of jobs. They get you 80% of the way there, but sometimes you need one more task that doesn't fit neatly into the existing output structure.

How It Works

  • Add a custom task — Define an additional output step on top of the agent's standard jobs. Think of it as a mini Content Agent embedded within your Enrich or Qualify workflow.
  • Same context, extra output — The custom task inherits all the same context the agent already has — Library, Playbook, runtime data — so it reasons with the full picture.
  • Flexible use cases — Generate a custom summary, extract a specific data point, produce a talking point, or create any additional output your workflow needs.

No need to chain a separate Content Agent for that last mile. Custom Tasks keep everything in one agent run.

Jan 2026

Qualification Question Weights

Set High, Medium, or Low weights on individual qualification questions, plus instant qualifiers and disqualifiers for more granular scoring control.
Agents

Previously, all qualification questions contributed equally to scoring. Now you can weight each question to reflect how much it actually matters to your qualification criteria.

Weight Levels

  • High — Must-have criteria. These questions have the most influence on the overall score.
  • Medium — Important but not decisive. Contributes meaningfully without dominating.
  • Low — Nice-to-have signals. Adds nuance without swinging the score.

Instant Qualifiers & Disqualifiers

  • Instant Qualifier — If this question is answered positively, the lead is immediately marked as qualified regardless of other scores.
  • Instant Disqualifier — If this question is answered positively, the lead is immediately disqualified. Use for hard deal-breakers like wrong industry or company size.

Combined with the existing per-question reasoning and confidence scores, weights give you much more precise control over how leads are scored and routed.

Dec 2025

Brand Voice

Define reusable brand voices with core traits, guiding principles, tonal qualities, and key terms. Apply across Sequence and Content Agents.
AgentsOptimization

Key Features

  • Reusable Brand Voices — Establish and store brand voices in the Styles section for deployment across Sequence and Content Agents.
  • Structured Voice Controls — Define core traits, guiding principles, tonal qualities to use and avoid, key terms, and any additional context relevant to the brand.
  • Easy Iteration — Brand voices support ongoing refinement through dedicated sections covering personality, writing rules, audience considerations, and tone.
  • Agent-Level Application — Once selected within an agent, brand voice automatically applies across all generated content with transparent credit usage tracking.
  • Reduced Prompt Overhead — Eliminates repetitive custom instructions and keeps brand consistency centralized and easy to manage.
Dec 2025

Smart CTAs

Dynamic CTA selection based on recipient profile. Define a CTA library with "when to use" logic, training examples, and hyperlinks.
Agents

Rather than using fixed CTAs per email, the system automatically selects the most appropriate offer based on recipient characteristics like persona, role, and company type during content generation.

Key Features

  • Dynamic CTA Selection — CTAs are matched during runtime based on recipient profile data, rather than being predetermined for each email.
  • CTA Library — Establish reusable CTAs within the Writing Styles section, each linked to a specific offering and guidance about appropriate usage scenarios.
  • "When to Use" Logic — Define conditions that determine when a CTA is most relevant, providing the agent with selection guidelines.
  • Training Examples and Hyperlinks — Provide examples controlling CTA presentation, along with optional hyperlinks and custom display text.
  • Agent-Level Control — Choose between auto mode (agent picks from full library) or manual mode (agent restricted to a specific CTA).