Agent
Discoverability
Agent Discoverability Research Lab

How AI agents choose
and use your product

Our research lab tracks the Plugin and Connector ecosystem across ChatGPT, Claude, Muse and Grok every day and publishes what it finds: live data, original reports and the method behind them.
TrackingChatGPTClaudeMuseGrokGeminiSoonCopilotSoon

Tracked daily since 1 December 2025

15,369integrations tracked daily
7,649ChatGPT Plugins
7,617Claude Connectors
59Muse Connectors
44Grok Connectors

Trusted by teams at

Statistamonday.comDirectBookerLucidMend.ioManpowerGroup

Presented at

Explore the research

The live data behind every report.

Our research isn’t only reports. Open the same tracking data yourself.

Ecosystem tracker

Every ChatGPT Plugin, Claude Connector and Muse Connector, with daily growth and category breakdowns.

Category leaderboards

Who ChatGPT and Claude pick first in each category, ranked and tracked over time.

Discovery scores

How often agents find and recommend a product on their own, by category and platform.

Published reports

Our latest published report.

Oct 17, 2026 report

Agent Tool Optimization: the new practice of getting your product discovered and used by agents

AEO optimizes for the citation in an answer. ATO optimizes for your tool being found, picked and called. In this research we explain the fundamental differences and mechanisms between optimizing for answers, ads, and the tools agents call (connectors and plugins), and how to improve your Plugin/Connector discovery.. An early look at both stores, built from our daily tracking: growth, the category map, capabilities, and what it means for getting your product discovered.

10 min read

Read the report
~15,300integrations live across ChatGPT and Claude, up from about 1,600 in May
~321Ktools indexed across the two stores, up from about 30K in May
80%of Claude action prompts end with a connector shown or invoked
30%of ChatGPT action prompts end with a plugin shown or invoked

Store sizes are live, tracked daily. Prompt figures are from our Research Lab, 12 Sep to 12 Oct 2026, free accounts in the US. New reports are added here as we publish them.

ResearchResearch on ChatGPT Plugins and connector stores

Research report

The State of the ChatGPT Plugin and Claude Connector Stores

What 1,597 live integrations reveal about the new agent distribution channel

May 24, 202619 min read
How we measure

Observed behaviour, not self-reported claims.

Observed, not claimed

Every result comes from watching what an assistant surfaces and selects for a prompt. Nothing is taken from a vendor’s own description of its integration.

Repeated and dated

Prompts run every day across phrasings and sessions, and every number carries its date. Assistants change weekly; a number without a date is worthless.

Honest about the evidence

Each report says which way a signal moves and how thin the evidence is, and we publish the methodology alongside the findings.

Read the full methodology
Research lab updates

Follow what the lab finds.

One email when the lab publishes a new report or dataset. No fixed schedule.