Proof of concept

Air Defence Training Simulator

A generic operational trainer. A domain expert teaches the system real scenarios by talking through them; trainees then ask for the practice they want in plain language, and the system builds an exercise from the captured expertise — then debriefs them against the expert’s own reasoning.

How this works

Three stages, in order — each one needs the one before it. Open any of them to pick up where you left off.

  1. Set up a system to train on

    1 ready

    Describe how a system behaves, build its console from screenshots of the real one, and teach it operational scenarios by talking them through. As many systems as you need, side by side.

  2. Put a trainee through it

    13 runs

    The trainee picks a system and asks for the practice they want in their own words. A matching scenario is found and an exercise built around it, then flown on the console.

  3. Review how they did

    8 debriefed

    Every run a trainee has flown: the score, the trend across sessions, and the full record of each decision alongside the debrief written against the expert's own reasoning.

What the system holds right now

Read live on every visit, so it never reports a state the app is not in.

Simulated systems
1 — 1 ready for trainingEuro_AD
Knowledge base
1 approvedCounted across every system. A scenario only ever matches within its own.
Simulated consoles
1 of 1
Storage
Git — committed to the repositoryScenarios and training runs alike are committed as files.
AI engine
claude-opus-5