Forgefy joins your planning meetings, transcribes them live, and extracts product requirements as you talk. Its AI agent pipeline turns the conversation into a build-ready blueprint, then generates your app no specs, no handoff, no lost context
No reviews yetBe the first to leave a review for Forgefy
Maker
📌
Hey Product Hunt! 👋
Every app starts as a conversation and then that conversation dies in meeting notes, half-written specs, and Jira tickets nobody reads. By the time anything gets built, half the context is gone.
Forgefy closes that gap. It sits in your planning meeting and:
🎙️ Transcribes in real time live speech-to-text as your team talks
🧠 Extracts requirements as you speak a multi-agent AI pipeline (4 specialized Claude agents) pulls out features, entities, flows, and constraints from the conversation
📐 Aggregates a build-ready blueprint everything discussed becomes a structured spec, streamed to your screen over WebSocket while the meeting is still happening
🚀 Builds the app the blueprint feeds a build agent that generates the actual application, multi-platform
The magic moment: you finish a 30-minute planning call and the blueprint of your app is already on screen.
Would love your feedback especially on what you'd want the blueprint to capture that meetings usually lose. Ask me anything! 🙏
Report
One thing I'd love to see is the ability to flag or pause during the meeting if we want to clarify something before the AI locks it into the blueprint. Sometimes requirements get said ambiguously and we need a quick confirmation moment instead of having to edit the generated spec later.
Report
Maker
@asiyeenisoczbn This is a great point, and honestly one we've been going back and forth on internally. Right now the flow is optimized for momentum, but you're right that it can bite you when a requirement gets phrased ambiguously and only shows up as a problem once you're editing the generated spec. A lightweight "flag this, ask me" moment mid-meeting feels like the right fix rather than a full pause/resume system. Appreciate you flagging it, added to the roadmap list
Report
Tried it on a quick planning call with a coworker and was honestly impressed it pulled out actual user stories without me babysitting it. The blueprint output was surprisingly detailed, basically what i'd write myself after an hour of cleanup.
Report
Maker
@glhaneb63 Okay this comment made my day. That hour of cleanup you're describing is precisely the tax we built this to remove, so hearing it actually held up on a real planning call with a coworker (not just a demo) is the best kind of feedback. Thanks for giving it a real test run
Report
The live transcription picking up requirements as we talked was honestly pretty smooth, and I was surprised how usable the generated blueprint was right out of the gate.
One thing I'd love to see is the ability to flag or pause during the meeting if we want to clarify something before the AI locks it into the blueprint. Sometimes requirements get said ambiguously and we need a quick confirmation moment instead of having to edit the generated spec later.
@asiyeenisoczbn This is a great point, and honestly one we've been going back and forth on internally. Right now the flow is optimized for momentum, but you're right that it can bite you when a requirement gets phrased ambiguously and only shows up as a problem once you're editing the generated spec. A lightweight "flag this, ask me" moment mid-meeting feels like the right fix rather than a full pause/resume system. Appreciate you flagging it, added to the roadmap list
Tried it on a quick planning call with a coworker and was honestly impressed it pulled out actual user stories without me babysitting it. The blueprint output was surprisingly detailed, basically what i'd write myself after an hour of cleanup.
@glhaneb63 Okay this comment made my day. That hour of cleanup you're describing is precisely the tax we built this to remove, so hearing it actually held up on a real planning call with a coworker (not just a demo) is the best kind of feedback. Thanks for giving it a real test run
The live transcription picking up requirements as we talked was honestly pretty smooth, and I was surprised how usable the generated blueprint was right out of the gate.