Connect Sync GTM to Your AI Client

Outcome: a working MCP connection, verified with a call that costs nothing, and two habits that stop a session spending more than you meant.

Surface: MCP server
Level: Beginner
Uses: check_credits
Credits: 0 — check_credits is free
Prerequisite: A Sync GTM account

What you are connecting

The Sync GTM MCP server exposes the same data actions the app runs — profile enrichment, email finding and verification, company research, signal detection — as tools your AI client can call directly.

Once connected, you stop switching tabs. You ask for a contact’s details in the same conversation where you are drafting the email, and the client fetches them mid-sentence.

The difference from the app is what you give up and what you gain. You give up repeatability — there is no table, so nothing re-runs on a schedule. You gain immediacy, which is the right trade for one-off rep work and the wrong trade for a recurring list build. Automated Outbound is where the repeatable version lives.

Steps

Install the server in your client

Setup differs per client, and each has its own page: MCP server setup covers Claude Code, the Claude app, Claude Cowork, Cursor, Codex, ChatGPT, Gemini, OpenClaw and Manus.

Follow the page for yours and come back here. Sign-in is through the browser — there is no API key to paste.

Verify with a free call

Once the client reports the server as connected, ask for your balance:


How many SyncGTM credits do I have left?

This calls check_credits, which takes no arguments and costs nothing. A number coming back means the connection, the authentication and the tool routing all work.

If the client answers from memory instead of calling the tool, say so explicitly:


Use the SyncGTM check_credits tool to get my current balance.

Confirm which tools are available


List the SyncGTM tools you have access to, grouped by what they do.

You should see roughly 50, across person enrichment, person search, LinkedIn profiles and posts, company enrichment and search, jobs, website scraping, traffic and ads, and local business listings.

The full catalogue with credit costs per tool is at Tools — worth skimming once, because knowing that two tools return overlapping data at different prices is what makes lesson 06 make sense.

Run one real lookup

Pick a LinkedIn profile URL you already know and ask for the basics:


Enrich this LinkedIn profile and give me their current title and company:
<paste a profile URL>

Read the output against what you already know about that person. You are checking that the data is right before you trust it on people you do not know.

Two habits, from the start

Bracket anything expensive with a credit check.


Check my credits, enrich the first 10 profiles from this list, then check my
credits again and tell me what that cost.

You learn the real per-row cost of a workflow on a batch small enough not to matter.

State volumes explicitly. “Enrich everyone in this list” against a pasted block of 400 rows is a request an AI client will carry out enthusiastically. “The first 20” is a request you can afford to get wrong.

These two habits cover most of what goes wrong for individual reps. If you are rolling MCP out to a team, budgets need enforcing rather than suggesting — that is AI for RevOps, lesson 03.

Check your work

Check	How
Server connected	Client lists SyncGTM tools
Auth working	`check_credits` returns a number
Tools reachable	A profile enrichment returns real data
Data is trustworthy	The result matches someone you already know

Where this breaks

If the client answers a data question without calling a tool, it is guessing from training data — and profile details in particular will be confidently out of date. Name the tool explicitly when in doubt, and treat any answer that arrives suspiciously fast as unverified. A tool call takes a moment; a hallucination is instant.

A connection that worked yesterday and fails today is almost always an expired browser session rather than a broken install. Re-authenticate through your client’s connector settings before reinstalling anything.

Further automation

Once the connection works, the workflows in this course can be saved as reusable skill files so you do not re-type the prompt chain each time — see GTM agents for the installable versions, and check there before writing your own.

Next lesson

02 — Your first contact card, chaining profile enrichment, work email lookup and verification into one contact you can actually reach, for about 2 to 3 credits.

Reference for this lesson: MCP server setup, Tools, Check Credits.