AI PRD generator
Generate a complete PRD. In four passes.
A guided 4-step flow with a multi-pass generator. Coach you through the brief, scan your existing context, draft the spec, and ship it. Each step adds the things your draft was missing before you knew they were missing.
Before & after
From a blank page to a brief that's ready.
Before Thinkr
- A blank doc and a one-line brief — you start writing and hope the gaps surface later.
- Scope, edge cases, and risks get found in review, or after launch.
- Every PRD starts from scratch, ignoring the specs and decisions already in your tools.
- Quality swings with who wrote it and how much time they had.
With Thinkr
- A brief scored on six dimensions before you write a line.
- Gaps, edge cases, and risks surfaced as you draft — not after.
- Each draft builds on the related PRDs, APIs, and risks already in your workspace.
- A consistent quality bar on every spec, in far less writing time.
Coach the brief
The Coach walks you through the brief in plain language, scoring readiness before you spend a generation pass on an incomplete spec.
- A readiness gate scores six dimensions of the brief
- Conversational refinement updates the working brief in place
- Brief revisions saved automatically
Pull in context
Before generation runs, Thinkr assembles the context your draft needs — from your Brain, from ClickUp, from prior PRDs — so the spec builds on what already exists.
- Context scan via ClickUp: prior PRDs, tech docs, API specs
- Pull documents from your Brain (workspace knowledge base)
- Citations preserved in the final draft
Draft in four passes
A multi-pass generator builds the spec — initial structure, then a gap-repair pass, then a self-critique, then a strengthen pass. The result reads like something a senior PM wrote, not a model output.
- Four passes: structure → gap-repair → self-critique → strengthen
- Structured user stories: a specific persona, a MoSCoW priority, and their own acceptance criteria
- A dedicated Edge Cases section, plus ≥2 Gherkin scenarios per flow (a happy path and an error path)
- Metrics that connect: each event maps to a success metric, and every metric gets an owner, threshold, and rollback trigger
- Structured non-functional requirements — including accessibility (WCAG) and localisation — with targets and owners
- Personas, typed dependencies, owned risks, real launch phasing, and a References section
- Rigorous mode: mark unknowns as [TBD] instead of inventing a plausible metric, name, or link
- Inline edits: improve writing, fix errors, apply a critique suggestion
- Draft completeness scored against a built-in rubric, dimension by dimension
- Version history with a diff viewer at every pass
Publish and ship
Publishing freezes the spec — Coach, Context, and Draft go read-only, so what you shipped stays what you shipped. From there, one Save & Share action sends it wherever it goes next.
- Save & Share: publish, or download as PDF or Markdown
- Send the spec into ClickUp as tracked work
- Turn the PRD into a proposal deck in one step
- Keep the finished spec as reusable context in your Brain
How it fits the loop