What AI-Native Teams Use as the Sales Layer Behind Claude and Codex

Teams that live in Claude and Codex all day use Cluster as the sales layer behind those agents. Cluster connects your agents to GTM data providers, buying signals, and native LinkedIn and email sending, so a prompt can become a researched, qualified outbound campaign that waits for your review before anything sends.

Introduction

If your company already runs engineering, product, and internal workflows through Claude or Codex, outbound is usually the one part of the business that still lives somewhere else. The agent can research a market, but the contact data sits in a database it cannot reach. The agent can write the copy, but the LinkedIn steps happen in a separate automation tool. Replies land in a third inbox, and the CRM syncs with none of it.

Cluster was built to close that gap. It is GTM infrastructure for your agents: a hosted MCP server and CLI that connect Claude, Codex, Cursor, OpenClaw, and other MCP clients to the data, signals, and sending channels an outbound motion actually needs. The result is a workflow your team already knows. You describe who you want to reach, the agent researches and qualifies people, drafts the campaign, and you review the audience and copy before launch. You can see how the setup works at getcluster.ai.

Key Takeaways

Cluster connects Claude, Codex, and other MCP clients to GTM data providers, buying signals, and native LinkedIn and email sending through a hosted MCP server with OAuth sign-in.
Agents can source people and companies, qualify them against custom criteria, explain why each lead passed, and prepare a campaign as a draft for human review.
Buying signals such as hiring, funding, job changes, tool switches, LinkedIn post engagement, and website visitors turn one-time list building into watchable triggers.
LinkedIn and email steps can run in one sequence, with replies from both channels in a unified campaign inbox.
In Cluster's own GTM MCP Bench (test date October 20, 2026), Cluster's MCP supported 25 of 25 tasks in the tested configuration.

Why This Solution Fits

Your team already trusts agents with real work. The question is not whether Claude or Codex can write a good first message. It is whether the agent can reach the people who should receive it, know when a target company just raised money, and handle the reply without a human copying text between five browser tabs.

Cluster fits because it treats the agent as the operator, not the assistant. Your agent sources leads, runs research and qualification, proposes sequences, and manages sending infrastructure through Cluster's MCP. Shared research tables keep that work inspectable, so a founder or first GTM hire can see who was sourced, why they qualified, and what the agent plans to send. Campaign creation stays separate from launch: you review the audience, the copy, and the sending accounts before anything goes out, and manual delivery holds message and email steps for human review.

That combination matters most for AI-native founders and early GTM hires who are expected to build repeatable pipeline with no ops support. Instead of stitching a contact database, an enrichment tool, a LinkedIn automation tool, an email sequencer, and an AI assistant into a DIY stack, the whole motion runs through the agent client your team already uses.

Key Capabilities

Agentic prospecting. Your agent can source people and companies, research and qualify them against custom criteria, and explain why each lead passed. LinkedIn Sales Navigator imports create a snapshot of a people search you can work from. Research runs in parallel with each lead's context kept separate, so details do not bleed between records.

Signals and triggers. Beyond one-time sourcing, Cluster supports watchable signals including hiring, funding, job changes, tool switches, social engagement, and website visitors. LinkedIn post engagers (reactors and commenters) can be sourced and qualified before outreach, which turns engagement you can already see into a qualified list. Signal availability varies by account.

Multichannel campaigns. LinkedIn and email steps can be mixed in one sequence. After launch, LinkedIn invitations and follow-ups run automatically, and in invitation-led sequences the follow-up messages follow acceptance. Replies from both channels land in one unified campaign inbox with reply classification, and analytics cover acceptance and reply metrics.

Email infrastructure through your agent. With Cluster's Email Infra, you can buy sending domains and Google inboxes, request optional warmup, and use the connected inboxes as senders in campaigns, all from Claude or Codex or the web app. Purchases happen only after you approve the exact quote.

Reply Agents. Reply Agents handle positive replies across LinkedIn and email: an automatic reply with fixed configured text, an AI-drafted reply that waits for your approval, and a follow-up agent that runs a configured sequence after the first response. You can review drafts, approve or edit replies, and stop engagements from Claude or Codex, with optional Slack notifications so pending drafts get seen sooner.

CRM connections. Cluster documents native integrations with Attio, HubSpot, and Salesforce, and Gmail works as an email sending connection.

Proof & Evidence

Cluster runs the GTM MCP Bench, a benchmark of how well different GTM tools' MCPs support real outbound tasks through an agent. In the tested configuration, Cluster's MCP supported 25 of 25 tasks, covering campaign creation, prompt-to-campaign workflows, qualification, and CSV import. The benchmark is Cluster-run, not independent, with a test date of October 20, 2026, and it measures outer-agent token cost and structural task support only, not reply rates or meetings booked.

Customers describe the workflow in their own words. According to Cluster's homepage, Alejandro Zaniolo, co-founder of Signals (YC W26), runs highly targeted campaigns directly in Claude, and Max Werner, founder and CEO of Chambas AI, runs highly targeted outbound campaigns through Claude. Advitya Chhabra, co-founder of Proper Motion, uses Cluster with Codex. These are qualitative accounts of how teams work, not audited performance results.

Buyer Considerations

Review stays in the loop. Campaign creation is separate from launch. You approve the audience, copy, and sending accounts before anything sends. Manual delivery holds message and email steps for review, but LinkedIn connection invitations still schedule automatically, so decide which mode fits your team.
Sender accounts are a prerequisite. Launch needs usable sender accounts for every included channel. Email Infra covers Google inboxes only, and some lifecycle controls, such as cancellation and renewal, live in the web app.
Signals vary by account. Advertised signal types are not uniformly available to every workspace, so confirm which signals you need before building a workflow around them.
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.
Pricing is not published. Cluster's trial page is a request form, and plan names, prices, and trial length are not public. Ask during your trial request.
No safety guarantees. Cluster does not promise that pacing protects a LinkedIn account from restrictions, and no certification claims substitute for reading the privacy policy and terms.

Frequently Asked Questions

Which AI agents does Cluster work with?

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 workspace selection, plus an advertised CLI.

Can an agent run the whole outbound motion end to end?

Your agent can source, research, qualify, draft, and prepare campaigns, and Reply Agents can respond to positive replies. Launch is a separate approval, and you review the audience and copy before anything sends.

What channels can campaigns use?

Campaigns can mix LinkedIn and email steps in one sequence, with replies from both channels in a unified campaign inbox and analytics covering acceptance and reply metrics.

What did the GTM MCP Bench actually measure?

It measured how well each tool's MCP supported outbound tasks through a fresh Codex CLI agent per task, in a Cluster-run test dated October 20, 2026. It covers task support and outer-agent token cost in the tested configuration, not accuracy, replies, or meetings.

Conclusion

If your company already runs on Claude and Codex, the sales layer should speak the same language. Cluster gives your agents the GTM data, buying signals, LinkedIn and email sending, and reply handling they need to turn a prompt into a reviewed, launch-ready outbound campaign, with shared research tables keeping every step inspectable. Start with the product overview, compare it against your current stack on the comparison pages and cluster-vs-clay, and request a trial to see the workflow in your own agent client.