02 — Learning Programme
6 entriesStructured study supporting the engineering work. The programme is organised by capability rather than by course provider: the courses are inputs to the experiment, not its output.
- 01 — Agentic AI Moving from conversational use of language models to systems that reason through tasks, use tools, plan work and evaluate their own intermediate results. 1 entry
- 02 — Model Context Protocol Investigating MCP as a controlled interface between Claude and external tools, data and services, including permissions and human approval. 1 entry
- 03 — Tools & Retrieval Giving the system access to relevant evidence through tool calling, retrieval, embeddings and evidence-grounded context while preserving provenance. 1 entry
- 04 — Agent Orchestration Exploring how multi-step and potentially multi-agent workflows should be structured, controlled and evaluated. 1 entry
- 05 — LLMOps Moving from prototype behaviour toward an observable, testable and operationally defensible AI system. 1 entry
- 06 — Evaluation & Acceptance Defining requirements, acceptance criteria, evaluation methods and measurable evidence that the system actually improves the target workflow. 1 entry