Gong MCP: Official, Read-Only, and Summaries Only
Yes — Gong has an official MCP server. Gong announced MCP support in October 2025 and now ships a first-party server that lets MCP-compatible AI tools query account and deal intelligence straight from Gong. The catch is what it hands back: it’s read-only, and it returns AI-generated summaries, not the raw call transcripts or message bodies underneath them.
The other thing worth knowing before you set it up: an MCP server hands Gong to an AI inside a conversation you start. It’s a doorway, not a worker. Nothing watches Gong for you, nothing fires when a deal goes quiet or a competitor gets mentioned on a call, and nothing runs while the chat is closed. Here’s exactly what the Gong MCP does, how to turn it on, where it stops — and what to use when you want Gong work that runs on its own.
For the wider picture, every app with an MCP server is listed by category.
What the Gong MCP server does
Model Context Protocol (MCP) is the open standard that lets an AI client — Claude, ChatGPT, Cursor, and others — talk to an outside app through a shared interface. Gong’s official server exposes three tools:
ask_account— answers questions about a specific account, grounded in that account’s Gong activity.ask_deal— answers questions about a CRM deal, pulling from the calls and emails tied to it.generate_brief— produces a structured summary across categories (risk, next steps, sentiment) for a deal or account.
The important nuance: every response is a synthesized insight, not the underlying material. Gong’s own docs are explicit that raw call transcripts, message bodies, and activity lists are not returned over MCP — you get the AI’s take on the data, not the data itself. It’s read-only, so nothing you do through MCP edits a deal, logs a note, or changes a call record in Gong.
A Gong tech admin sets it up first. Registration can be automatic (clients like ChatGPT’s Gong plugin connect through a normal Gong sign-in, no credentials to copy) or manual (Gong issues a client ID and secret for one specific client). Access can be scoped personal (your own permissions) or shared (organization-wide). It’s available on any Gong plan, but you need a paid Gong seat (collaborators can’t use it), and every question consumes Gong credits based on how many calls and emails it analyzes.
How to set up the Gong MCP server
- A tech admin opens Admin center → Settings → Ecosystem → MCP, clicks the MCP servers tab, then New MCP server integration. They name it, enter the AI client’s redirect URI, and pick a registration type and personal or shared access (neither can be changed later).
- In your AI client, connect Gong. In ChatGPT that’s the Gong plugin under Plugins; Microsoft Copilot has a Gong connector too. In Claude, add a custom connector under Customize → Connectors with the server URL
https://mcp.gong.io/mcp(plus the client ID and secret for a manual integration). Then sign in to Gong. - Confirm
ask_account,ask_deal, andgenerate_briefappear as available tools, then start a chat and ask it a question about a live deal or account.
There’s no local server to host and no code to write for the official path. It’s an admin-configured connection, not a developer project.
Where the Gong MCP stops
None of this is a knock on MCP — it’s just the shape of the protocol, plus Gong’s own read-only design choice. The limits that show up the moment you want more than a conversation:
- It only works inside a chat you start. Close the window and nothing happens. The AI doesn’t watch Gong; it waits for you to ask.
- No triggers. A deal’s risk score dropping, a competitor mention on a call, a stretch of silence on an account — none of these can start anything through MCP. There’s no “when this happens in Gong, do that.”
- It’s one app at a time, and read-only. The Gong MCP knows Gong, and only in summary form. Getting a deal-risk signal into Slack, a CRM task, and a follow-up email means wiring up separate MCP servers for each and hoping your client can chain them — and none of them can write back to Gong itself.
- Your admin owns the setup, and every answer costs credits. Registration type and the personal-vs-shared access decision are on your tech admin and can’t be changed after saving; each question draws down Gong credits.
So the Gong MCP is a solid way to ask about an account or deal and get a grounded summary back. It is not a way to make Gong-driven work run — to have something happen on a schedule or in reaction to a call, across the other tools a deal touches.
Running Gong work that doesn’t need a chat open
That “run on its own, across apps” gap is exactly where Carly fits. Carly connects to Gong with an API key your tech admin creates in Gong’s admin center (no OAuth client to register, no MCP integration to maintain), and to thousands of other apps. The difference from MCP is the important part: Carly’s workflows are triggered and scheduled, so Gong-driven work happens whether or not anyone has a chat window open.
A few things that MCP can’t do but a Carly workflow can:
- When a call gets tagged with a competitor mention → notify the
#competitive-intelSlack channel and log the deal to a tracking sheet, automatically. - Every Monday → summarize accounts with no Gong activity in 10 days and email the list to each account owner.
- When a deal’s sentiment turns negative → create a CRM task for the AE and draft a check-in email for approval.
The non-AI steps — the moving, matching, and routing between apps — are free, the Zapier-style backbone of the workflow. The AI steps (drafting, summarizing, deciding) start at $35/month. You describe the outcome in plain language and Carly wires up the Gong connection and everything downstream.
If you just want to interrogate an account or deal from a chat, Gong’s official MCP server is the right tool, as long as you have a paid Gong seat and credits to spend. If you want Gong signals to actually trigger things — on an event, on a schedule, across every app a deal touches — that’s the job MCP wasn’t built for, and it’s the one Carly was.
FAQ
Does Gong have an official MCP server?
Yes. Gong announced MCP support in October 2025 and ships a first-party, read-only MCP server with three tools — ask_account, ask_deal, and generate_brief — for querying account and deal intelligence.
Does the Gong MCP server return call transcripts? No. Gong’s documentation states plainly that raw data — call transcripts, message bodies, and activity lists — is not returned over MCP. You get AI-generated summaries, not the source material.
Can the Gong MCP trigger automations? No. It’s read-only and request/response inside an AI chat — nothing runs when the conversation is closed. For event- or schedule-driven Gong work across apps, you need a workflow tool like Carly rather than an MCP server.
Can I connect Gong to AI without coding or hosting a server? Yes. You don’t have to touch OAuth or MCP setup at all. Carly connects to Gong for you and lets you build the automation in plain language — describe what you want to happen and it wires up Gong and the other apps involved, with no server to configure and no code to write.
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
