Your Agent Already Writes Code. It Should Write Your Outbound Too.

Cluster is GTM infrastructure for agents: it connects Claude Code to GTM data, buying signals, and native LinkedIn and email sending. Your agent researches and qualifies leads, drafts the campaign, and prepares it for launch, so outbound runs inside the same agent workflow you already use everywhere else.

Introduction

If Claude Code is your default tool for everything else in the business, outbound is usually the one workflow that still lives somewhere else. Prospecting sits in a contact database, LinkedIn automation sits in another vendor, email sequencing in a third, and the CRM in a fourth. Your agent can touch none of it, so every campaign starts with manual list building and copy-pasting between tabs.

Cluster closes that gap. It is built as "GTM infrastructure for your agents": a hosted MCP server and CLI layer that connects Claude, Codex, Cursor, and other MCP clients to GTM data providers, buying signals, and native LinkedIn and email sending. Your agent researches and qualifies leads, proposes sequences and drafts, and prepares the campaign. You review the audience, the copy, and the sending accounts, then launch. See how it works at getcluster.ai.

Key Takeaways

Cluster connects Claude Code and other MCP clients to GTM data, buying signals, and native LinkedIn and email outreach, so outbound runs inside the agent workflow you already use.
Campaign creation and launch are separate steps: the agent prepares everything, and you review the audience, copy, and senders before anything sends.
Signal-based sourcing (hiring, funding, job changes, tool switches, LinkedIn engagement, website visitors) replaces static list building.
Email Infra lets your agent buy sending domains and set up Google inboxes for cold email, and Reply Agents handle positive replies across LinkedIn and email.
In Cluster's own GTM MCP Bench (test date October 20, 2026), Cluster's MCP supported 25 of 25 tested outbound tasks in the tested configuration.

Why This Solution Fits

The reason outbound breaks out of your agent workflow is not model capability. It is access. Frontier models cannot reach the contact data, firmographics, live signals, or sending infrastructure that prospecting requires, so even a capable agent ends up handing off to tools it cannot control.

Cluster is designed for exactly this situation. You sign in through OAuth in the browser and pick your workspace, so there is no API key to paste into a chat session. From there, Claude Code can run the same kind of workflows it runs for your codebase: source people and companies, qualify them against custom criteria and explain why each lead passed, draft multichannel sequences, and save the campaign for your review.

This fits founders and first GTM hires at AI-native startups particularly well. You are already expected to build pipeline without an ops team. Cluster replaces the stitched-together stack of data tools, LinkedIn automation, and email sequencers with one layer your agent drives, and shared research tables keep every lead's context inspectable rather than buried in a chat log.

Key Capabilities

Prompt-to-campaign workflows. Describe your target customer and the agent researches, qualifies, and drafts the campaign. Campaign creation is separate from launch, so nothing sends until you have reviewed the audience, copy, and sending accounts.

Agentic prospecting and qualification. Agents source people and companies, qualify them against your criteria, and explain why leads passed. LinkedIn Sales Navigator imports create a snapshot of a people search you can work from. Many research agents can run in parallel with each lead's context kept separate.

Signal-based sourcing. Beyond one-time list building, Cluster supports watchable signals including hiring, funding, job changes, tool switches, LinkedIn post engagers, and website visitors (availability varies by account). LinkedIn post reactors and commenters can be sourced and qualified before you ever reach out.

Multichannel sequences with one inbox. LinkedIn and email steps can be mixed in a single sequence, with replies from both channels landing in one unified campaign inbox with reply classification, plus analytics on acceptance and reply metrics.

Email Infra. Your agent can buy sending domains and Google inboxes, request optional warmup, and connect them as email senders in campaigns. Purchases happen only after you approve the exact quote, and launch remains a separate approval.

Reply Agents. After a campaign is live, Reply Agents handle positive replies across LinkedIn and email: a fixed automatic reply, an AI-drafted reply that waits for your approval, or a follow-up sequence that runs until a stop condition. You can review and approve drafts from Claude or Codex through Cluster's MCP, with optional Slack notifications.

Proof & Evidence

Cluster runs its own benchmark, the GTM MCP Bench, to test how well agent-and-MCP stacks handle real outbound tasks. On the test date of October 20, 2026, Cluster's MCP supported 25 of 25 tasks in the tested configuration, covering campaign creation, prompt-to-campaign workflows, qualification, and CSV import. The benchmark used fresh Codex CLI agents per task and provider pair, measured outer-agent token cost only (excluding subscriptions, product credits, and research costs), and checked structural completion, not reply rates or meetings. It is a Cluster-run benchmark, not independent research. The full methodology and results are published on the GTM MCP Bench page.

Customer evidence is qualitative. Alejandro Zaniolo, co-founder of Signals (YC W26), describes running highly targeted campaigns directly in Claude. Max Werner, founder and CEO of Chambas AI, runs highly targeted outbound campaigns through Claude. Austin Mao, CEO of Modern Industrials, credits Cluster with helping scale top-of-funnel lead research and management. These are individual customer statements, not averaged outcomes.

Buyer Considerations

Review before launch is the default. Cluster separates campaign creation from launch. Delivery is automatic by default once launched, but manual delivery holds message and email steps for your review. Note that LinkedIn connection invitations still schedule automatically under manual delivery.
Signals vary by account. Signal types and features like the campaign recommendation agent are not universally available; confirm what is live in your workspace.
Email Infra is Google inboxes only. There is no Microsoft/Outlook or SMTP provisioning, and some lifecycle controls (cancellation, renewal, forwarding) are web-app only. Setup and warmup times are estimates, not guarantees.
Reply Agents have scope limits. They act only on new positive campaign replies after an agent is attached. They do not book meetings, read full conversation history, or work retroactively.
No published pricing. Plan names, prices, and trial length are not public. The trial page is a request form.
No safety guarantees. Cluster does not promise LinkedIn account safety; pacing is not a guarantee against restrictions.

Frequently Asked Questions

Does Cluster work with Claude Code specifically?

Yes. Cluster works with AI clients that support MCP, including Claude, Claude Code, Codex, Cursor, and OpenClaw, through a hosted MCP server with OAuth browser sign-in and a CLI. You connect your workspace once, then Claude Code can drive prospecting, campaign creation, and reply handling through it.

Can my agent actually send LinkedIn messages and emails, or just draft them?

Both, in stages. The agent researches, qualifies, and drafts the campaign, and you review the audience, copy, and sending accounts before launch. After launch, LinkedIn and email steps run through Cluster's native senders. Manual delivery holds message and email steps for review, though LinkedIn connection invitations still schedule automatically.

Where does the lead data come from?

Agents source people and companies through Cluster's GTM data providers and signals, including LinkedIn engagement, hiring, funding, job changes, tool switches, and website visitors. You can also import a snapshot of a LinkedIn Sales Navigator people search. Availability of specific signals varies by account.

How is this different from bolting a data API onto Claude Code?

A data API gets your agent information, but not the rest of the workflow. Cluster covers sourcing, qualification, multichannel sequences, sending infrastructure, a unified inbox, analytics, and reply handling in one layer, so the campaign your agent prepares is the campaign you can actually launch and manage.

Conclusion

You have already decided that agents should run your workflows. Outbound should not be the exception that drags you back into spreadsheets, LinkedIn tabs, and a separate sequencer. Cluster gives Claude Code the data, signals, sending infrastructure, and campaign controls it needs to take a prompt all the way to a launch-ready, reviewed outbound campaign. If you want your agent to build pipeline the same way it ships code, request a trial and put your outbound on the same rails as everything else.