I built an audio/video-to-content tool because I was tired of the manual grind
The Problem:
I spent years helping clients repurpose podcasts and coaching calls into LinkedIn posts, tweets, and newsletters. Every. Single. Time:
- Transcribe (boring but easy)
- Pull out the good quotes (hard)
- Write in their voice (harder)
- Get them to approve it (nightmare)
The Aha Moment:
What if AI could do the extraction part AND learn the client's voice over time? Like, actually remember that they hate "utilize" and prefer "use"?
What I Built:
PostPrism - upload audio, video or text, get 8 content pieces in 90 seconds:
- LinkedIn post
- 5 tweet hooks
- Twitter thread (5-7 tweets)
- Carousel outline
- Newsletter intro
- Key takeaways
- Show notes
- Blog post (800-1200 words)
The Twist:
The AI has "Context Memory" - it learns from every edit you make. Edit something once, it remembers for next time.
The Numbers (honest):
- Built in 4 weeks (solo dev)
- 0 paying customers yet (just launched)
- 3 beta users testing
- ~$49/month fixed costs (Supabase + Vercel)
- Targeting 90%+ gross margin
The Stack:
- Next.js 14 + Tailwind
- Supabase (auth, db, storage)
- Deepgram (transcription)
- Gemini 2.0 Flash (content generation)
- Claude (fallback)
- Stripe (payments)
What's Next:
10 beta users → validate quality → demo video → Product Hunt
The Ask:
If you're a coach, podcaster, or run a content agency, I'd love early feedback. Happy to give you extended trial access.
Also curious: what makes you actually convert from trial to paid on a SaaS tool?
