I’m building BraveHuman, an AI-based roleplay app designed to help people practice complex, real-world conversations.
Right now, the core experience revolves around what I call “stories” — structured, multi-step roleplays where each scenario connects to the next.
Instead of a single isolated prompt like “practice an interview,” users move through a narrative thread. For example:
– You’re having a casual coffee chat with a colleague – Later you’re invited to a pub hangout – On the way, your boss calls about an urgent client issue
The goal is to simulate the unpredictability and context-switching of real conversations, rather than a single static Q&A exchange.
Each step involves interacting with an AI character and receiving feedback before moving forward.
I’m currently exploring two directions:
1. A more structured training architecture (e.g., breaking practice into explicit “training methods” instead of pure narrative flow)
2. A personalized scenario builder where users can generate custom practice environments (interviews, presentations, networking, etc.)
I’d really value feedback on a few things:
– Does narrative continuity meaningfully improve practice over isolated prompts?
– How would you approach measuring improvement in communication quality?
– If you were building this, would you double down on story-driven roleplay or move toward fully customizable scenarios?
Still early, but people are using it and iterating has been interesting.
Would appreciate critique on assumptions, architecture direction, or blind spots.