Skip to content
The TAI Labs community is now on Skool
TAI Labs
Sign inBook a scoping call

For members

Use TAI Labs from the chat you already have open

Add one MCP server and Claude or ChatGPT can reach your account: build and run the agents that do your web work, catch up on your rooms, watch what the runs are costing you, and rewrite your CV from what you actually shipped — without opening the app.

  • Build an agent from a template

    “Show me the TAI Labs agent templates for research, then set one up to check my three biggest competitors every Monday morning and run it once now so I can see what it comes back with.”

  • What did my agents do this week?

    “Go through my TAI Labs agent runs from the last seven days: what came back, which ones failed and why, and what it all cost me in credits.”

  • Set one loose behind a login

    “Which sites can my TAI Labs agents already sign in to? Build me one that checks my orders there every morning and uses that saved login.”

  • Catch up on my rooms

    “Catch me up on my TAI Labs rooms — what is new in the feed, which conversations mention me, and draft a reply to the one that actually needs one.”

  • Post what I built

    “Take what my agent found this morning, write it up as a post for the TAI Labs feed, show me the draft, and post it once I say go.”

  • Find the right person and introduce me

    “Find someone in my TAI Labs rooms who has actually shipped an agent in production, open a DM, and introduce me in two sentences.”

Add https://tailabs.ai/mcp as an MCP server and sign in with your TAI Labs account. Your data stays yours — the assistant only sees what your account can already reach.

Build on Tai

An MCP server, an API and a Slack app

The same tools are available to your own stack: generate a programme from a topic or your own documents, read lessons, run quizzes and track progress.

  • MCP server

    Sign in once and your assistant gets your agents, their runs, your rooms, your credits and your CV — 32 tools, all scoped to your account.

    https://tailabs.ai/mcp

  • Training API

    Generate a full programme from a topic or your own source material. JSON in, structured modules out.

    POST https://tailabs.ai/api/public/course-outline

  • Slack app

    Training built inside your workspace, in the channel where the work is being discussed.

    @Tai training on incident response for new engineers, 8 lessons

Training API

One POST, one structured programme

Send a topic — and optionally your own source material — and get modules, lessons with measurable objectives and hands-on practice, quiz topics and a final project. Authenticate with your Tai API key.

curl -X POST https://tailabs.ai/api/public/course-outline \
  -H "Authorization: Bearer $TAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "topic": "Incident response for new engineers",
    "audience": "engineers in their first 90 days",
    "level": "beginner",
    "lessons": 8
  }'
{
  "outline": {
    "title": "...",
    "outcomes": ["..."],
    "modules": [
      { "title": "...", "summary": "...",
        "lessons": [{ "title": "...", "minutes": 12,
                      "objective": "...", "practice": "..." }] }
    ],
    "assessment": { "quizTopics": ["..."], "project": "..." }
  }
}

MCP

Connect Tai to Claude, ChatGPT, Cursor or Codex

One OAuth-protected endpoint gives your assistant your agents, your runs, your rooms and your credits — signed in as you, with exactly your permissions. The step-by-step for every client lives on one page, so it never drifts from what the server actually serves.

https://tailabs.ai/mcpConnect your assistant

Slack

Build a programme in the channel where the work happens

Mention Tai with what you need and it replies in the thread with the full outline. Requests are signature-verified and answered by the same engine as the API.

@Tai training on incident response for new engineers, 8 lessons, beginner