Version 1.0, 2026-10-05.

How HumanAI works

HumanAI is a workplace for data and engineering teams who work with AI agents on their customers' systems. Agents do the heavy lifting and people decide.

1. Each person has their own environment, isolated and set up with the project's tools. It keeps working with the laptop closed. You use it from the browser, on a computer or a phone.
2. Agents work under your identity. You connect the agents you want (Claude, Codex, Kimi) and register your own access to the customer's systems. When an agent opens a pull request or comments on a ticket, it is recorded under your name. Nobody uses someone else's credentials.
3. The team coordinates in one place. Need someone to approve, review or decide? Send them a request; they get it in the cockpit and on their phone, answer, and the work carries on from there. Production approvals go to whoever has the authority. Requests that are already resolved close automatically.
4. What one person learns, everyone keeps. Traps and recipes are recorded, reviewed and made available to the agents and the rest of the team.
5. The day starts with what matters. "Today" gathers what needs you: requests, code reviews, failed builds, changed tickets. Pipeline alerts open work for whoever has to act.
6. Meetings that leave a record (optional). With participants' consent, the meeting app transcribes, identifies speakers and prepares the summary and decisions.

Where the data stays. Servers are in the EU. From customers' data platforms we bring metadata and aggregates, not data rows. Content an agent needs goes to the provider the customer authorised.