Clay MCP is the layer that brings Clay's data and workflows into the AI tools your team already works in, so reps act on Clay without opening Clay.
MCP, the Model Context Protocol, is an open standard for connecting an AI assistant to an outside system. Clay's MCP server uses it to expose 150+ data providers, AI research agents, and your team's prebuilt workflows inside ChatGPT, Claude, and Codex. The result reframes what Clay is for a rep: not a separate platform to learn, but a set of capabilities they reach from the chat window they already live in. This guide explains what Clay MCP is, how it works, where it runs, and how reps and RevOps use it together.
What Clay MCP actually is
Clay MCP is a connector, not a new app to learn. Historically Clay was a workflow platform built for ops: deep, but it took time to learn and was easy to burn credits in, which kept most reps out of it. MCP changes the access model. The same data and workflows now surface inside the assistant, so a rep types a request in plain language and Clay answers in the conversation. RevOps still builds the logic in Clay; reps just consume it where they already are. Clay's own framing says it plainly: ops-built workflows, consumable by reps.
Auto-plays a request flowing from the AI tool through MCP to Clay and back. Click a layer to hold it and read what it is.
Your prompt
You type: "Find the VP of Sales at stripe.com and research the account"
Clay MCP is a connector layer: the AI tool stays the interface, MCP is the bridge, and Clay's providers, research agents, and prebuilt workflows are what it exposes.
How Clay MCP works
MCP makes Clay a tool the assistant can call, the same way it calls any other connector. When a rep asks the assistant to find people, research a company, or run a workflow, the assistant routes that request through Clay's MCP server, Clay executes it against the right provider or agent, and the structured result returns into the conversation as an interactive view. Nothing is copied between tabs and nothing requires the rep to open Clay. The assistant handles the language; Clay handles the data and the work.
Where Clay MCP works
Clay MCP runs in the major AI assistants, with the same core capabilities in each. The connection differs slightly by tool, but the capabilities are consistent: find and verify people, research accounts, draft outreach, and run your team's Functions. New accounts get 500 free Clay credits on first connect.
Clay MCP across AI tools
| AI tool | How you connect | What you get |
|---|---|---|
| ChatGPT | Type @Clay (browser) or /Clay (desktop), then authenticate | Find, enrich, research, and draft in an interactive view |
| Claude | Add the Clay connector from the Claude Connectors page (Enterprise: an admin enables it) | Ask in natural language; the connector activates automatically |
| Codex | Connect Clay's MCP server | Run Clay data and workflows from the coding assistant |
What reps can do with it
The connector gives reps four jobs that used to span a dozen tabs. Each is something a rep asks for in the chat, and the context carries from one to the next:
- Find and verify contacts at a target account: With verified emails and work history, returned as an interactive view.
- Research the account before reaching out: Hiring trends, tech stack, funding, and leadership changes, gathered by an AI research agent.
- Write personalized outreach: Using the research already in the conversation, so the message is built on what the account cares about.
- Prep for a meeting: A one-screen brief on the account and the people in it, ready before the call.
Auto-plays the old multi-tab way against one Clay MCP conversation. Tap a side to hold it; Replay restarts.
The value of Clay MCP is collapsing the multi-tab research-and-write scramble into a single conversation where context carries between steps.
How RevOps and reps share one system
The point of Clay MCP is that RevOps builds once and reps execute everywhere. The old split was the whole problem: ops built workflows in one place, reps worked in another, and the data lived in a third. MCP puts all three in the same loop. RevOps configures access, sets credit limits, and packages vetted workflows; reps invoke those workflows from their AI tool; usage flows back to the admin to monitor. Reps get consistent, governed output without learning the platform, and ops keeps control of quality and spend.
Auto-plays the ops-builds, rep-runs, admin-monitors loop. Click a node to hold it and read it; click the held node to resume.
The loop repeats: build, run, monitor, refine.
RevOps / admin
Configure access, set credit limits per rep, and package a vetted workflow as a Function with an actionable name.
Clay MCP unifies the ops-builds / reps-execute split into one governed loop: admins package and control workflows, reps consume them in their AI tool, and usage stays monitored.
Functions: your workflows, available in the chat
A Function is the unit that makes rep output consistent. It is a reusable Clay workflow an admin builds once, then turns on with Enable for MCP and gives an actionable name. After that, every rep calls the same vetted workflow by name from ChatGPT, Claude, or Codex, instead of each writing their own prompt and getting their own answer. That is how a team gets one "Account brief" or "Company enrichment" output everyone trusts, rather than a dozen freehand versions. Behind that one-line prompt sits everything MCP exposes: the 150+ provider marketplace, AI research agents, and your team's own packaged Functions.
Auto-plays ad-hoc prompts vs one packaged Function. Click a lane to hold it and inspect it.
Packaging a workflow as an MCP Function turns ad-hoc rep prompting into one vetted workflow every rep calls by name, so output stays consistent and reps never open Clay.
When to use Clay MCP versus the Clay platform
Clay MCP is for rep-scale, in-the-moment work; the platform is for volume and automation. Reach for the connector when a rep is handling 1 to 20 contacts, doing exploratory research, or writing individual emails. Move into the Clay platform for 20+ contacts, complex automated workflows, and deep CRM syncs. The signal that a workflow has outgrown the chat is repetition: if a rep types the same prompt over and over, that workflow should be packaged as a Function or built as a table, not retyped every time.
Auto-sweeps task size across the connector vs Clay-platform threshold. Drag the slider or click Replay to inspect.
Clay MCP (in the chat)
do it in the prompt- Exploratory research
- Ad-hoc planning
- Individual emails
Repeating a prompt? If you are typing the same prompt over and over, package it as a Function or build it as a table.
Admins set per-rep credit limits and monitor usage from the MCP section of the Clay side nav.
Clay MCP is for rep-scale, ad-hoc work (1 to 20 contacts); volume, automation, and CRM sync belong in the Clay platform, and any repeated prompt should become a Function.
When the whole team works this way, adoption shows up in the usage.
enrichments OpenAI's sales team ran in Clay, a sign of team-wide adoption once the data lives where reps work
Read the full storyWhere to start
Start by connecting one AI tool and running a single real prospecting task end to end. Connect Clay in ChatGPT or Claude, then find a contact, verify the email, research the account, and draft the outreach in one conversation to feel the difference. For the full setup and the prompt patterns, read our guide to using Clay in ChatGPT with MCP. When a workflow proves itself, have RevOps package it as a Function so the whole team runs the same one.
Next step: walk through the full setup and prompt patterns in How to Use Clay in ChatGPT (MCP), then bring the workflow that proves itself back to RevOps to package as a Function.
Outcomes teams report after putting Clay in their reps' hands
What teams report after giving reps Clay-powered data and research
The same pattern shows up when a team consolidates its data tools into Clay and hands reps the context up front.
“We consolidated three vendors into Clay and started enriching data points that didn't exist in any traditional database. Our reps went from starting every conversation cold to knowing exactly who to call and what to say.”