- Description
- Curriculum
Context Engineering in the SDLC — course description
The context supply chain
AI can draft your requirements, test cases and automation code in minutes. But it drafts from whatever context you hand it — and most of that context was written for humans, not machines. Agents rarely fail at random; they fail on context that’s missing, stale, or contradictory. Almost all of that is fixable before an agent ever runs.
This half-day, presenter-led workshop follows your context through five stages — Raw Material, Refinery, Meter, Warehouse, Quality Control — mixing the strategic case with hands-on technique. You’ll audit your real sources, rewrite requirements and documents so AI retrieval can actually find them, cut token bloat without losing meaning, and put ownership and freshness in place so readiness doesn’t decay. Live before/after rewrites throughout, plus an honest look at where a traditional AI copilot helps and where it stops short.
You’ll take away six ready-to-use templates: a landscape audit worksheet, an AI-ready user-story template, worked token-count examples, a do/don’t reference sheet, the AI-readiness scorecard, and per-role checklists.
Format: ~3 hours · 5 stages · 12 modules · product-agnostic principles
-
3
What AI sees when it looks at your workGet an honest picture of the sources feeding your AI today, and learn a repeatable four-question audit you can run instantly. You'll leave able to score any source, decide what to fix first — and what to delete outright.
-
4
Turn vague requirements into something buildable
Learn how to prepare requirements, documents and diagrams for AI usage. You'll be able to write acceptance criteria an agent can automate, structure any wiki page so its content is actually found, and spot contradictions before they reach your AI Copilot.
-
5
Why less context gives better results
Discover why feeding AI more usually makes it worse, and cheaper context often makes it better. You'll be able to estimate what any input costs, cut it by half without losing meaning and calculate the trade-offs.
-
6
How to keep context up-to-date
Stop your work from decaying. You'll leave with ownership that survives post-AI, a context lifecycle that is deeply embedded in your processes and six failure patterns you'll recognise in your own team immediately.
-
7
Learn to score context, and know what to fix first
Make context quality visible — a number your team can track and your team can act on. You'll leave able to measure the quality any context source, see where you stand, and know how to take action.
What you'll learn
- Map and audit your context landscape against four tests: findable, readable, trustworthy, necessary
- Write requirements and acceptance criteria any AI agent can build and test against
- Structure documents for retrieval — chunking, headings, metadata
- Cut token cost while raising AI accuracy
- Establish a single source of truth, with ownership and freshness that hold
- Score readiness and act on it, role by role
No prior experience needed
Who it's for Business, functional and technical analysts · product owners · scrum masters · software engineers · test engineers · IT leadership. No coding required.
