Collaboration with AI for developers.

The engagement.

Organisation

Ministry of Defence (MOD)

Audience

Developers

Focus

Collaboration with AI

Think with AI.

The curriculum begins with Bloom’s Collaboration Framework and the Auditability Line: distinguish work you can check against a source from decisions whose consequences you own. AI can inform a decision; the human retains judgement.

Participants practise defining purpose, success and scope before asking AI to work. Deeper decisions add evidence, unknowns, reversibility and a named human decision maker.

From judgement to a working system.

The curriculum develops a collaboration surface, adds tools and controls, then connects it to a software interface. The stages below summarise the learning material.

1. Establish the collaboration surface.

   Organise notes, references and decisions so both the developer and AI can use them. Practise finding sources, connecting knowledge and keeping a readable Git trail.

2. Give AI useful tools and human boundaries.

   Create reusable skills for capturing knowledge, recording decisions and recalling information. Separate read actions from writes, then practise reviewing a proposed change before accepting it.

3. Make the checks enforceable.

   Build guards, validation and a controlled command path. Exercise blocked actions, repair invalid input, inspect the work model and test the rules rather than relying on a prompt to enforce them.

4. Design, build and review the interface.

   Develop a task tracker and knowledge interface. The material covers data storage, read and write paths, UX, wireframes, interface states and the review loop: inspect the change, run it, then decide.

Practice that can be inspected.

Labs in the curriculum include writing a purpose note, linking a decision to its evidence, building a capture skill, testing an approval guard and connecting an inbox to a human acceptance action.