Loading image: AI Product Discovery Playbook - Verlin Labs AI guide cover image…AI Product Discovery Playbook
Shipping an AI feature without discovery is guessing with GPUs. This playbook gives product managers a repeatable path: map the job-to-be-done, score AI fit, prototype with off-the-shelf models, define guardrails, and cut scope to a testable MVP.
Verified learning path
This guide connects to live Verlin Labs programs - not generic AI content. Apply these frameworks in a cohort with mentor feedback and a capstone demo.
Job-first, model-second
Write the user story without mentioning AI: "When I ___, I need ___ so that ___." Only then ask whether an LLM removes friction cheaper than rules, search, or human ops. Many "AI features" are better served by better UX or cleaner data.
Interview five users on the current workaround - screenshots, spreadsheets, copy-paste habits reveal the real job.
AI fit scorecard
Rate the task on ambiguity, error tolerance, data availability, and regulatory exposure. High ambiguity + low error tolerance = expensive to ship safely. Document which failures are acceptable in v1 and who approves overrides.
- Summarisation and drafting: strong fit when sources are provided.
- Autonomous decisions on money or health: defer or human-in-the-loop.
- Personalisation: needs privacy review and consent flows upfront.
Prototype before the PRD hardens
Spend two days testing prompts on real inputs with design partners. Capture failure modes in a spreadsheet - categories beat anecdotes in stakeholder reviews. Your PRD should list non-goals explicitly to prevent scope creep into "full autopilot."
MVP cuts that still teach
Ship assistive flows before autonomous ones: suggest, do not send. Log edits users make to AI drafts - that signal prioritises the next iteration. Pair quantitative metrics (time saved, override rate) with qualitative trust interviews.
Key takeaway
Discover the job, score AI fit honestly, prototype on real inputs, and ship assistive MVPs with logged overrides. PMs who document non-goals ship faster and safer than teams chasing demo magic.
