Skip to content
product-skill
product-skill Templates Fill-in skeletons

PRD · AI-codegen PRD · Strategy · OKR · Business Plan

Pick one, fill it, and lead with the decision — not the structure.

Five fill-in skeletons, each copyable as Markdown. The moment a feature is LLM-powered, switch to the AI-codegen PRD. Non-Goals and open questions are mandatory, not optional.

Template · 10-section PRD

PRD — [Feature name]

1TL;DR + decision requested
2Problem (+ evidence)
3Goals — business & user
4Non-Goalsmandatory
5User Stories / JTBD
6UX (flows, mocks)
7Narrative (press-release)
8Success Metrics (primary + guardrails)
9Technical + open questions (tag confidence)mandatory
10Milestones (phases, not dates)
+Dependencies (incl. instrumentation cost)

Kevin Yien (Square), popularized via Lenny Rachitsky — not a "ChatPRD standard"

Above the fold

The three moves that carry a good template.

01

Pick ONE, fill it, lead with the decision.

Don't lead with the structure. A template earns its place by making the decision requested obvious on the first line — everything else supports it.

02

Non-Goals are mandatory, not optional.

Non-Goals must name excluded segments AND the cheaper alternative you considered NOT building. The open-questions list, tagged with confidence, is mandatory too.

03

The moment it's LLM-powered, switch templates.

Move to the AI-codegen PRD: specify WHAT, never HOW. For LLM features the golden-prompt set + eval rubric IS the spec.

Template 1 · the workhorse

PRD — 10-section.

PRD — 10-section

Kevin Yien (Square) · popularized via Lenny Rachitsky

  • 1TL;DR + decision requested
  • 2Problem (+ evidence)
  • 3Goals — business & user
  • 4Non-Goals — what we're explicitly NOT building
  • 5User Stories / JTBD
  • 6UX (flows, mocks)
  • 7Narrative (press-release / customer story)
  • 8Success Metrics (primary + guardrails)
  • 9Technical considerations + open questions tagged with confidence
  • 10Milestones (phases, not hard dates)
  • +Dependencies — incl. cross-team instrumentation cost (the analytics/events another team must ship for you to measure success)

§4 Non-Goals + the open-questions list are mandatory. Non-Goals must name excluded segments (e.g. single-player/single-seat users are NOT the target) and the cheaper alternative you considered NOT building (e.g. improve existing email before building an in-app center).

For regulated/sensitive features, spec regulated-content handling — masking, audit logging, sensitive-data-on-shared-screen rules. Multi-context calibration: B2C leans habit-formation hooks; early-stage PRDs over-scope by default — and don't over-build v1 to match Slack/Linear's mature surface.

Scoping discipline · Milestones §10

Dollarize Cost of Delay as $/month and recurring LTV impact (not a one-time number) so the bet ranks on the right magnitude — then ship the narrowest slice first (one IdP before a SAML + OIDC + SCIM + directory-sync sprawl). A narrow slice that ships beats a broad slice that slips.

Template 2 · when it's LLM-powered

PRD for AI codegen — 6-section.

ChatPRD format. Specify WHAT, never HOW: describe behavior, inputs/outputs, constraints, and testable acceptance criteria; let the coding agent choose implementation.

AI-codegen PRD — 6-block

ChatPRD format

  • 1Overview / goal
  • 2User stories
  • 3Functional requirements (observable behavior)
  • 4Non-functional constraints (latency, cost, safety)
  • 5Acceptance criteria
  • 6Out-of-scope

Attach the golden-prompt set + eval rubric (→ §AI-Native). For LLM features, that IS the spec.

Template 3 · direction + execution in one doc

Strategy doc — "Strategy Blocks."

Strategy Blocks

Chandra Janakiraman (Lenny's, 2025)

  • 1Winning-aspiration headline
  • 2~3 pillars + explicit non-goals
  • 3~3 HMW questions
  • 44-dim bet rubric (incl. uniqueness / defensibility)
  • 55-phase timeline

Frame each bet as a hypothesis — "We believe [action] will result in [outcome], as measured by [metric]" — with an inline DHM assessment and an explicit "What we're NOT doing (and why)" block. Keep small-s execution and big-S direction in one doc; cite as a named framework, not settled canon.

Template 4 · the fill-in

OKR — objective, key results, guardrails.

OKR fill-in

baseline → target + confidence

  • OObjective — qualitative, inspiring
  • KRKR1–3, each with baseline → target + confidence
  • GGuardrail-metrics slot — what must not regress
  • IKey Initiatives = the bets, NOT the results

The most common OKR error is listing initiatives as key results. Key Initiatives are the bets; Key Results are the outcomes those bets aim to move.

Template 5 · when the financial spine matters

Product Business Plan — the Aumayr spine.

Reach for it on B2B-industrial / P&L-owning PM work. The full fill-in skeleton lives in PRO — this is the spine to copy and expand.

Product Business Plan

Aumayr

  • 1Market
  • 2Positioning
  • 3Strategy
  • 4Financials (contribution margin, break-even point)
  • 5Roadmap
  • 6KPIs

Full fill-in skeleton lives in PRO MODULES → Strategy & Roadmap . Reach for it where the financial spine matters.