Capstone · Playbook
Capstone: Assemble your AI-in-HR playbook
You now have five work products. Put them together as a single playbook that you could show to a manager or use as the starting point for a real programme.
| Section | From | Contents |
|---|---|---|
| 1. Opportunity map | Module 1 | Three scored use cases with canvases, KPIs and guardrails. |
| 2. Content system | Module 2 | Voice card, prompt templates, quality rubric, and the onboarding communication plan. |
| 3. HR assistant | Module 3 | Scope card, instructions, knowledge inventory, test log, change log, maintenance plan. |
| 4. Data story | Module 4 | One-page brief or slides, methods note, prompt log. |
| 5. Automation | Module 5 | Specification, evidence of a working Zap, test log, runbook. |
Add two things
- A one-page summary using the “what, so what, now what” structure: what you built, what it shows about where AI can help HR, and what you recommend doing next.
- A 30-60-90 day plan for yourself. First 30 days: use the content system on real, low-risk work and record time saved. Days 31 to 60: pilot the assistant or the automation with a small group, with approval. Days 61 to 90: report results against your baselines and propose the next use case.
Final self-assessment
- Can I explain to a non-specialist how a language model produces text and why it can be wrong?
- Can I place any proposed AI use on the risk ladder and name the controls it needs?
- Can I write a prompt with context, role, action, format, and tone and constraints, and ground it in a source?
- Can I verify an AI-drafted document against its sources using a rubric?
- Can I scope, instruct, test and maintain an assistant?
- Can I calculate core HR metrics correctly and turn them into a costed, honest recommendation?
- Can I specify, build, test and document an automation with a human checkpoint?
- Do I know which legal themes apply and when to involve counsel?
Draft your answer here. It saves automatically and appears in My playbook.