Version 1.0, 2026-10-05.

AI use in HumanAI

What AI does here. HumanAI puts AI agents (Anthropic's Claude, OpenAI's Codex, Moonshot AI's Kimi) to work alongside people: they read code and documentation, propose changes, prepare analyses, tickets and documents, and run tasks in systems the user can access. The cockpit always shows which agent and model you are talking to.

What is sent to the provider. Your request, the files and code the session opens, and the results of the tools the agent uses; no more than the task needs. Providers and their safeguards are on the subprocessors page. Each customer chooses the providers its projects may use, and the platform blocks the others. Kimi (Moonshot AI, China) is blocked in projects with EU customers' personal data.

What we do not do. HumanAI does not train its own models on customer content, and does not use AI to make automated decisions about people.

A person reviews and decides. Users can make the agent plan before acting and approve or refuse each action with effects; production changes need approval from a person with authority; whoever delivers AI-assisted work reviews it and is accountable for it.

Known limitations. Agents can make up facts, misread a request or use outdated information. Check figures, names and claims before using them outside the platform.

Work produced with AI. Documents and code produced in the platform are reviewed by a person who takes editorial responsibility before delivery. Anyone who publishes AI-generated text to inform the public on matters of public interest, without that human review and editorial responsibility, must disclose that it was AI-generated, as the EU AI Act (art. 50) requires. Our own content follows the same rule.

Contact: rpad@rpad.pt