The four-pillar structure — Design · Build · Evaluate · Govern— and the framing that anchors this course both come from Dr Bloomfield's “Agentic AI: Design, Build & Govern” programme at Cambridge Judge Business School Executive Education. The REMIT framework, the Agent Canvas used throughout the Design module, the executive-readable tone, and the through-line that autonomy is earned, not granted are all his.
Credits
This course stands on other people's shoulders. Here is who, and for what.
Built this interactive companion: the lesson content beyond the original slides, the scenario builder, the canvas tooling, the diagrams, the playgrounds, the cost and context calculators, and the underlying plumbing. Additional writing across the Build, Evaluate, and Govern modules — including material on context windows, prompt caching, MCP, evaluation strategy, and current regulatory practice (EU AI Act, NIST AI RMF, Anthropic's Responsible Scaling Policy).
Much of the course was drafted, refined, and illustrated in close collaboration with Claude — both as a writing partner on lesson copy and examples, and as a coding partner on the interactive components, playgrounds, and this site itself. Every word has been reviewed by a human before publication, but it would be dishonest to pretend the work was done without LLM assistance. It seemed fitting, given the subject.
Source materials and references
Where the course leans on specific outside research or frameworks:
- Liu et al. — Lost in the Middle: How Language Models Use Long Contexts (2023). arxiv.org/abs/2307.03172
- Chroma Research — Context Rot: How Input Length Degrades LLM Performance (2025). trychroma.com/research/context-rot
- Anthropic — Responsible Scaling Policy v3.0, Model Context Protocol (MCP), and Claude model documentation.
- NIST — AI Risk Management Framework (AI RMF 1.0).
- European Commission — EU AI Act, as adopted 2024 and entering force in stages through 2026.
Tools this was built with
Next.js App Router, React, TypeScript, Tailwind, shadcn/ui, MDX, Recharts, the Vercel AI SDK, and the Vercel AI Gateway. Hosted on Vercel. In dev, the tutor and canvas-critique agents run against a local Ollama instance.
If you are reusing material from this course in your own work — slides, internal training, blog posts — please credit Dr Mark Bloomfield for the underlying course design and Simon Elliston Ball for the interactive adaptation. If anything here helps you ship a safer, more thoughtful agent, that is the only KPI this course has.