Skip to main content
The PRD workflow converts canvas specifications into structured product requirements documents.

​
Generation flow

1

Canvas to context

The frontend reads all canvas nodes and edges, serializes them, and sends them to the generation API endpoint.
2

Streaming progress

The UI shows real-time progress as the PRD generates via SSE. A progress indicator shows the current stage (reading canvas, generating, finalizing).
3

Rendered output

The completed PRD is rendered as formatted Markdown with section navigation. Users can scroll through and review each section.

​
Refinement

The refinement interface is a conversational UI alongside the PRD. Users provide feedback on specific sections, and the AI updates the document. Each refinement:
  • Creates a new version
  • Preserves the conversation history
  • Shows a diff against the previous version

​
Version comparison

The version picker lets you compare any two PRD versions side by side. Changes are highlighted as additions, removals, and modifications.

​
Export options

FormatImplementationUse case
MarkdownRaw text downloadImport into docs or wikis
PDFPDFMake renderingSharing with stakeholders
Imagehtml-to-image captureEmbedding in Slack, presentations

​
Ticket extraction

The ticket extraction tool parses a PRD into discrete work items. Each ticket includes a title, description, and suggested priority. Extracted tickets can be pushed to Linear, Jira, or Notion through the integrations system.