The pilot program

Start with a 7-day sovereign AI pilot.

No cloud trial accounts, no toy data. The full platform installs on your server, inside your network, and works on your code — and the pilot ends with evidence your security team can examine.

Apply for the pilot

What the pilot includes

  • A 7-day, full-feature evaluation
  • Installed on your GPU server — inside your network
  • A local open-weight model, ready to work
  • The full coding agent with approvals and plan mode
  • The admin and governance console
  • Audit logs and the Agent Flight Recorder
  • A security review session with our engineers
  • Exported audit evidence that stays with you

The evaluation unlocks the full enterprise feature set — and cleanly self-destructs after seven days unless licensed.

Eight steps — from first call to evidence

  1. 01

    Discovery call

    We understand your environment, sector, and first use case.

  2. 02

    Hardware check

    We review your GPU server, network, and isolation requirements.

  3. 03

    Install on your server

    One command over SSH installs everything: hardware check, NVIDIA/CUDA, database, app, and web tier.

  4. 04

    Connect one test repository

    A repo you choose; nothing else is reachable.

  5. 05

    Run 3 approved coding tasks

    Real tasks from your backlog, with human approvals.

  6. 06

    Export the audit evidence

    The Flight Recorder log and signed manifests — in your hands.

  7. 07

    Security review

    A session with your security team on the architecture and the evidence.

  8. 08

    Commercial proposal

    If the evidence convinces you — we map the full deployment.

A decision built on evidence — not a slide deck.

Tell us your sector and environment, and we'll come back with the pilot plan and hardware requirements.

Apply for the pilot