Dating Coach
An AI trainer where you practice real dating conversations with AI characters — proof the pipeline scales to a different domain.
Quick facts
- Launched
- 2026-04
- Stack
- Flutter · FastAPI · LLM · subscription
- Audience
- Consumer
- Status
- Live
Why I built it
One product proves nothing — it could be a fluke. So I deliberately picked a second product in a completely different niche: not dream interpretation, but a conversation-heavy, character-driven trainer where people practice dating skills with AI characters.
Four chat modes. A character system with distinct personalities. Memory-driven personalization that makes the AI feel like it actually knows the user. Radically more complex than Dreambook’s content model. The question was simple: does the method survive a harder domain?
It did. The pipeline held. The same approach — decisions by a human, execution by the AI — scaled to a product that required a different kind of architecture, a different monetization model, and a completely different content strategy. That’s the validation. Two products, two domains, one pipeline.
What I learned
- Character-driven, conversation-heavy architecture is a fundamentally different design problem from Dreambook's content model. The AI pipeline handled the domain shift — the architecture decisions still needed a human.
- Onboarding redesign: an expert-led plan that shows users exactly what they'll practice outperforms a feature tour every time. People don't want to understand the product. They want to feel the value.
- Migrating monetization from credits to subscription mid-flight — without breaking active users — is the kind of problem that looks impossible until you map the backward-compat path first.
Proof
Screenshots and metrics — coming in Phase 6.
Want to see the product itself?