6 weeks · 2 hours per week
The Engineering AI Intensive
Recover engineering time across the whole lifecycle — not just the coding step.
A six-week programme for an intact engineering team. Not Claude training, and not another dev-tool license: six sessions spent finding where engineering time actually leaks — review backlogs, stale docs, manual QA, onboarding, ticket triage — and reclaiming it with Claude Code and Cowork. Run it for the whole team, because the value shows up in handoffs, not just in commits.
- Length
- 6 weeks · 2 hours per week
- Time commitment
- 2 hours a week for the whole team, plus homework
- Who it is for
- Engineering teams: developers, plus ideally a PM, QA lead, or tech writer in the room. The cross-functional angle is the point.
- Formats
- Whole-team workshop · Optional IC and lead/manager tracks
Outcomes
What changes by week six
Not a certificate. Six weeks of applied work you can still point at afterwards.
- ✓
A measured before-and-after on PR cycle time, onboarding time or review turnaround
- ✓
Team conventions for AI-assisted code: shared prompts, project setups, review guidelines
- ✓
Claude Code used for real delegated work, not autocomplete
- ✓
Cross-functional workflows that PM, support, docs and QA can run themselves
- ✓
A company-specific starter kit the team keeps using after the programme ends
The Syllabus
One skill a week. Nothing overlaps, nothing rushes.
Each week has a clear focus and a lab that ends with something finished on your own work. This is a working plan — it gets adjusted with you as the cohort progresses.
- 1 Week 1
Where AI fits beyond the IDE
The full lifecycle
Map planning, coding, review, testing, docs, deployment and support — and show where each person in the room already loses time. Introduce Claude Code as an agentic collaborator rather than autocomplete: delegating multi-step work across a whole repository, not one file at a time.
Lab work
Everyone finishes one real task that is not "write this function" — e.g. explain an unfamiliar module, or trace where a bug could originate.
- 2 Week 2
Agentic workflows & delegation
Briefs, not prompts
Multi-file refactors, running tests, fixing failures, opening PRs. How to write a task brief for an agent — context, constraints, definition of done — versus a traditional prompt, and the difference between chatting with AI and delegating a ticket to it.
Lab work
Delegate one real backlog ticket or chore — a dependency bump, a test coverage gap, a lint cleanup — end to end.
- 3 Week 3
Code review, quality & docs
Highest ROI, lowest trust risk
Drafting PR descriptions, reviewing diffs for risk and style, generating and updating docs and READMEs, explaining legacy code to new joiners. This is the week that converts the sceptics, because nothing about it feels risky.
Lab work
Use Claude to document or review one real PR from this week.
- 4 Week 4
Testing & QA
From spec to reproducible test
Generate test cases from requirements, hunt edge cases, turn bug reports into reproducible test scripts, and use Claude Cowork on file-and-data-heavy QA such as cross-checking data against specification documents.
Lab work
Generate a test suite or test plan for a real feature.
- 5 Week 5
Cowork & cross-functional workflows
From my IDE to our operations
The explicit shift from personal tooling to team operations: turning a stack of customer bug reports into a prioritized backlog, summarizing sprint retros, auditing folders of configs or docs for inconsistencies, generating onboarding material.
Lab work
Delegate a cross-functional, non-coding task to Cowork.
- 6 Week 6
Integration, governance & scaling it
Making it stick
Connect Claude to the tools the team already uses — GitHub or GitLab, Jira, Slack, Google Drive — through MCP and connectors. Team-wide conventions: shared prompt libraries, project setups, style guides for AI-generated code, review and security practice, and the metrics that prove ROI: cycle time, PR turnaround, onboarding time.
Lab work
Each person or pod presents a workflow they built and the time it saved, plus the company-specific starter kit.
Inside Every Session
Two hours, same rhythm every week
Total: 2h of structure (120 minutes), so there is always enough time to actually practise rather than watch.
- 10m
Recap
Metrics, wins and blockers from the week.
- 30m
Concept + demo
One idea, shown live on the team’s own repo.
- 45m
Team lab
Everyone works; coached in real time.
- 25m
Role breakout
IC and lead tracks split where useful.
- 10m
Commitments
One workflow to run before next week.
A baseline rather than a rule — batches per week and time splits can be changed to suit the group.
What you need
- → A real repository the team works in — examples come from your own codebase
- → Claude Code installed and authenticated for each participant
- → A rough baseline of PR cycle time, onboarding time or review turnaround
- → PM, QA or tech writer in the room if you have them
What’s included
- ✓ Six 2-hour sessions with the full team, remote or on-site
- ✓ A baseline measurement before Week 1 and a re-measure afterwards
- ✓ A company-specific starter kit: prompt library, project setup, review guidelines
- ✓ Optional compressed 2-day version, or an AI readiness audit as a lighter entry point
The Engineering AI Intensive — next cohort
Tell me whether you want 1:1 or a small group, and what your biggest bottleneck is right now. We will confirm timing, and flag anything worth doing before week one.