Clay MCP: How to Set It Up and Use It for Prospecting

How Clay Functions Work With MCP

Functions are useful when a team already has established GTM processes. Ops defines the inputs, enrichment steps, formulas, and outputs once. Reps call the finished workflow from their AI client.

• Function: Enterprise Account Brief
• Input: Company domain
• Workflow: Firmographics > funding > tech stack > hiring signals > ICP logic
• Output: Account brief + ICP score

A rep can simply ask:

Run Enterprise Account Brief for acme.com.

Custom Functions are created by your team and can be made available through MCP. Give them clear names that describe the outcome. A Function called Find contacts can be confused with Clay’s built-in people search, while Enterprise Buying Committee Finder is much more specific.

Functions do not create a separate MCP fee. Any credit usage comes from the Clay actions the Function runs.

What Clay MCP Can and Can’t Do

Clay MCP gives an AI client access to specific Clay search, enrichment, research, and Function tools. It does not give Claude, ChatGPT, or another MCP client full control over everything inside your Clay workspace.

How Much Does Clay MCP Cost?

Clay does not charge a separate fee for using MCP. The cost comes from the Clay credits used by the searches, enrichments, and Functions you run.

There are a few useful things to know before you connect it:

• You get 500 free Clay credits when you connect Clay to Claude for the first time. Clay says this applies whether you are creating a new Clay account or connecting an existing one.
• People and company searches can continue for free after those credits run out. Credit-consuming enrichments, such as finding an email address, require available Clay credits.
• Enrichment uses the same credit pricing as Clay itself. MCP does not add a surcharge.
• Functions use the credits consumed by the enrichment steps inside them. The Function itself does not add another fee.
• Team admins can cap MCP spend per user. Credit limits can be set from Settings > MCP users and adjusted individually.

For example, if a Function that finds an email and phone number costs 12 credits when it runs inside a Clay table, running the same Function through an MCP client costs the same 12 credits.

Once the initial 500 credits are gone, enrichment draws from your existing Clay plan. Clay’s pricing therefore depends much more on what you ask it to enrich than on the fact that you are using MCP. A simple people search and a multi-step enrichment Function can have very different credit usage.

Pros and Cons of Clay MCP

The main benefit of Clay MCP is running quick research and approved workflows without leaving your AI client. The main limit is that it does not replace Clay for bulk enrichment or automated work.

長所

• Quick for one-off prospecting: Research or enrich a few contacts without creating another Clay table. It is also an useful addition to your sales productivity tools.
• Fits into the AI tools reps already use: Clay can be used from Claude, ChatGPT, and other supported MCP clients, making quick research easier to fit into an existing workflow.
• Keeps repeatable workflows consistent: RevOps can build a Function once in Clay and let reps run the same process through natural language without recreating the enrichment logic.
• Gives admins useful controls: Teams can manage MCP users, enabled Functions, approved clients, and credit limits from Clay.

• Lets you refine results as you go: The interactive Clay view supports additional filtering and enrichment without forcing you to start the search again.

短所

• Not designed for large-scale work: Clay MCP is better suited to small, ad hoc tasks than bulk enrichment, large prospect lists, or recurring workflows.
• Enrichment still uses Clay credits: The initial 500 credits are useful for testing, but paid enrichment continues to draw from your Clay balance once those are used.
• Some workflows need setup first: If the action is not available through Clay’s standard MCP tools, RevOps may need to create and enable a Function before reps can use it.
• The experience varies by AI client: Clay works across several MCP clients, but the connection flow, invocation method, and way results appear can differ between Claude, ChatGPT, and other clients.

• Results still depend on the underlying data: MCP does not solve missing or unverifiable contact and company data. If Clay cannot find it reliably, the AI client cannot fill that gap.

Clay MCP Permissions and Admin Controls

From Settings > MCP users, admins can manage rep access, set credit limits, monitor usage, and control which Functions are available. Clay explains these controls in its MCP settings documentation.

Teams can also restrict which MCP clients users are allowed to connect. That is useful when a company wants Clay available in approved tools but not every compatible AI client.

Connecting Clay does not give the AI client unrestricted access to the whole workspace. Access depends on the user’s role and the tools or Functions the team has made available.

Common Clay MCP Problems and Fixes

Most Clay MCP problems trace back to three things: workspace access, remaining credits, or how a Function is set up. The fixes below cover the ones that come up most.

1. Clay Does Not Appear in Claude

Check that Clay is connected and that the user belongs to the correct Clay workspace. In managed Claude environments, an admin may need to make the connector available first.

2. Search Works but Enrichment Fails

Check the user’s remaining Clay credits and any MCP credit limit set by the workspace admin. A search can work even when a credit-consuming enrichment has been blocked.

3. A Function Does Not Appear

Open the Function in Clay and confirm that it is available for MCP. Also check that the Function produces a usable output and that its name does not closely overlap with one of Clay’s built-in tools.

4. Clay Finds the Wrong Company

Use the company domain instead of relying on the company name alone. Add role, seniority, location, or other filters where the request could match several companies or people.

5. Connected to the Wrong Workspace

Disconnect Clay from the AI client, accept the correct company workspace invite, then reconnect and select the right workspace.

6. A Custom MCP Client Returns invalid_grant

Custom clients using OAuth need to handle token refresh correctly. If a client keeps reusing an old refresh token, authentication can fail and the connection needs to be refreshed.

So, Is Clay MCP Worth Using?

Clay MCP is useful for finding the right accounts, identifying relevant contacts, and enriching them with the data you need before outreach. But prospecting data alone cannot tell you which messages actually resonate once those prospects start talking to your team.

That is where conversation data becomes useful. Clay does not have a direct integration with tl;dv, but tl;dv MCP can be connected to the same AI client and provide context from your recorded meetings.

Together, they can support a more useful loop:

Clay finds and enriches prospects → sales conversations happen → tl;dv captures and analyses those conversations → your team learns which segments, objections, needs, and messaging come up most often → Clay helps research and enrich the next set of relevant accounts and contacts.

That can also improve outreach. Instead of relying only on firmographic data, reps can use patterns from real customer conversations to understand which problems are worth leading with and what language prospects actually respond to.

If you want to bring conversation intelligence into that workflow, you can try tl;dv.