Skip to Content
Docs are being rebuilt — start at Introduction → How it works.
IntroductionWhat is TeamMate?

What is TeamMate?

TeamMate is the place where your team’s AI employees live. You give each one a name, a job, and the tools and knowledge it needs. Then you (and your team) talk to them, send them work, or let them run jobs on their own.

If you’ve used a chat tool like Slack and a process tool like Zapier, you can think of TeamMate as both — sitting on top of an AI that can read documents, write replies, fill spreadsheets, and call your other systems.

The one-paragraph version

A workspace holds your team and your agents. Each agent has a job description, the tools it can use, and the knowledge it can search. You talk to an agent in chat, send it work as a task that runs once, schedule it to run on a clock, or wire it into a workflow that chains multiple agents and external systems. Anything an agent does can be governed — approval steps, audit logs, role-based access — so you don’t lose oversight when you scale.

What real teams build with it

Concrete examples we see in production today:

  • HR question-and-answer. An agent grounded in your company’s HR policies and Qatar labour law. Employees ask “how many annual leave days do I have?” and get a cited answer in seconds. Read the HR & people-ops guides.
  • Sales follow-ups. A coordinator agent that drafts personalised post-meeting emails, books a follow-up calendar slot, and posts a summary to a Slack channel. See SDR that books meetings.
  • Translation between Arabic and English. A bilingual agent grounded in Qatar legal terminology, used by legal, HR, and partnerships teams.
  • Invoice intake and coding. Inbound invoice PDFs get parsed, coded against the chart of accounts, and posted into Odoo with a human approval step. See Invoice intake & coding.
  • Daily digests. A scheduled task that runs at 8 AM, summarises the last day’s tickets / sales pipeline / GitHub PRs, and emails the result to the relevant manager.
  • Reception by voice or chat. An agent that takes inbound phone calls (through Twilio or Vapi) or chats on your website, answering common questions and handing off to a human when it can’t.

The shared shape: a domain-specific agent plus the right knowledge plus a way for users to reach it.

What TeamMate is not

To save you time:

  • It’s not a chatbot builder. Yes, you can build chatbots in it. But the same product runs unattended scheduled jobs, autonomous workflows, and voice receptionists. Chat is one surface, not the only one.
  • It’s not “wrap an LLM in a UI”. Out of the box you get governance, knowledge retrieval, model fallback when a chosen model can’t handle a file type, approval gates, audit logs, and per-workspace credit accounting. The interesting parts are the parts that aren’t the LLM.
  • It’s not infrastructure-as-code for one customer at a time. Workspaces are first-class. One workspace per company, one customer per workspace. You pay per workspace; usage is tracked there.
  • It’s not lock-in. Agents can use OpenAI, Anthropic, DeepSeek, OpenRouter, or any OpenAI-compatible custom endpoint (your LiteLLM proxy or vLLM cluster). Knowledge bases use open formats. Chats and audit logs are exportable.

Where to next

  • Continue learning: How it works walks through the building blocks and how they connect, with a small diagram.
  • Just try it: Quick start (5 min) gets you signed in and chatting in under 5 minutes.
  • Show me real workflows: Guides — end-to-end recipes by industry and department.
IntroductionHow it works