The fastest way
to write a PRD.
Paste your customer interviews, tickets, and feedback. SpecForge clusters what matters, scores it by impact, and hands you a structured spec your team can ship from.
Everything you need.
Nothing invented.
From raw customer signal to a structured spec in minutes — not days.
Signal ingestion
Upload PDFs, paste transcripts, drop in support tickets. Everything chunked and embedded automatically. No formatting required.
AI clustering
Gemini 2.5 Flash groups recurring themes from hundreds of signal chunks into distinct, labelled opportunities.
Impact scoring
Each opportunity is ranked by frequency and severity. You see what actually matters to customers, not just what was mentioned last.
Structured PRD output
User stories, acceptance criteria, edge cases, and an agent-ready prompt — all Pydantic-validated. Export as markdown, hand to your coding agent.
Four steps.
One decision.
The pipeline runs automatically. You step in once — to approve which opportunities become spec — and the rest is handled.
# LangGraph pipeline
graph.add_node("ingest", ingest_node)
graph.add_node("cluster", cluster_node)
graph.add_node("score", score_node)
graph.add_node("review", human_checkpoint)
graph.add_node("generate", generate_node)
# runs only after PM approves
graph.add_conditional_edges(
"review", route_after_approval
)Paste your signal
Upload PDFs, paste interview notes or support tickets directly. No template to fill out, no fields to map. SpecForge handles the structure.
Let the pipeline run
Gemini 2.5 Flash clusters themes, scores each one by frequency and severity, and surfaces a ranked list of opportunities. Takes under 90 seconds.
Approve what matters
Review the ranked opportunities. Edit scores, rename themes, reject what's noise. One click to approve and send to generation. You stay in control.
Ship the spec
A structured PRD — user stories, acceptance criteria, edge cases, and an agent-ready Claude Code prompt. Export as markdown and hand it to your coding agent.