# What percentage of GTM tools can an AI agent reach? 59%, and less once you count the gate

> 59% of 1,252 GTM tools have an MCP server. Once the access gate is applied the reachable number drops again. The arithmetic, counted 2026-09-27.

*Markdown twin of the HTML page at the same path. Same content, no navigation, no styling, no scripts. Links below point at other twins. Site map for machines: [llms.txt](../llms.txt). The whole dataset: [directory.json](../data/directory.json).*

*Maintained by Andrew McGuire (https://andrewcmcguire.com), who also publishes https://gtmsignals.co and https://justsaid.ai.*

---
[Directory](../index.md) /
[Learn](index.md) / What the data says

**What the data says**

# What percentage of GTM tools can an AI agent actually reach?

**The short answer**

59% of the 1,252 go to market tools counted here have an MCP server of any kind. Applying the second filter, that you can actually get an account without a contract, 168 entries pass both tests, which is 13% of the directory.

## The funnel

| Filter | Entries left | Share of 1,252 |
|---|---|---|
| Every tool counted | 1,252 | 100% |
| Has a server of any kind | 741 | 59% |
| Server is first party | 201 | 16% |
| Server, and free or paid self serve | 168 | 13% |
| Server, first party, and free to start | 62 | 5% |
| And somebody here has actually run it | 1 | 0% |

Counted 2026-09-27 from directory.json. The last row is the honest floor of this whole exercise and it is published on the front page rather than buried here.

## Why the last two rows matter most

Every percentage above the last two is a documentation reading. It says a vendor published something and a check found it on a date. The bottom row says somebody ran the thing. This directory's bench tested count is 1, it stays there until Andrew personally runs a tool on a stated date, and no vendor can buy a change to it.

## The direction of travel

Do not read 59% as a permanent state. It is a reading taken on 2026-09-27 from 1,252 entries whose facts were pulled by hand on 2026-08-24, 2026-08-25, 2026-09-02, 2026-09-03, 2026-09-07, 2026-09-12, 2026-09-14. The point of stamping every number with a date is that the next reading is comparable to this one.

## Sources

- [The GTM MCP Directory, the counted data](../data.md) this site
- [The GTM MCP Directory, methodology](../methodology.md) this site

Every number on this page is generated from directory.json at build time and carries the date it was baked: 2026-09-27. Nothing is typed by hand, nothing is rounded, and nothing is estimated. The underlying data is [published in full](../data.md). Where the honest answer is a zero, the zero is printed.

## Related questions

- [How many GTM tools have MCP servers?](how-many-gtm-tools-have-mcp-servers.md)
- [How many of these GTM tools have actually been tested?](how-many-gtm-tools-are-bench-tested.md)
- [Which GTM tools can a solo operator use with an AI agent?](which-gtm-tools-can-a-solo-operator-use.md)
- [Which GTM tool categories are most usable by AI agents?](which-gtm-categories-are-most-agent-ready.md)

## In the directory

- [The data endpoint](../data.md)
- [The lists](../lists/index.md)