Consumer AI
kiddi.
A consumer app exploring how AI can help parents navigate everyday questions.
Context
Modern parenting is fragmented across dozens of tools, channels, and well-meaning experts. Information is everywhere, confidence is not. kiddi started from a simple question: what would a daily companion look like if it actually understood the family it was built for?
Founded inside the TUM.ai Entrepreneurship Lab, the project moves at the speed of a venture but with the discipline of a research environment. Early hypotheses were tested with real parents, not just stakeholders, and the product direction was sharpened around the few moments where AI changes the day-to-day, not just the long-term plan.
The prototype explored repeated, lightweight use and a way to retain useful context between conversations.
Contribution
- Translated the founding hypothesis into a clear product narrative, target user, and metric to validate.
- Ran early user research and concept tests with parents across different family setups and stages.
- Prototyped the AI-native experience end-to-end using Cursor, Claude, and Supabase — owning the build as much as the strategy.
- Defined the pre-beta scope and the validation gates needed to move from concept to invite-only release.
- Shaped the early commercial logic: monetization angles, GTM channels, and partnership opportunities.
Outcome
A validated pre-beta product with early user traction, sharpened positioning, and a clear path into invite-only release — built from concept to working prototype within a single lab cycle.
Reflection
The thinner the slice you ship, the louder the signal. Most of what felt essential on the whiteboard turned out to be optional; the moments that retained users were the ones that respected their time and assumed nothing about their household.