ZeroShame

ZeroShame was created to address the fact that 74% of young people experience difficulty accessing sex education due to conflicting online content and social taboos. The startup partnered with Personit to build a safe, reliable, and completely anonymous platform delivering intimate health guidance for Generation Z. The project began with discovery and design workshops to define the chatbot user flows, full deployment specifications, and bilingual architecture (Polish and English). The technical solution combines a modular app with Supabase for backend infrastructure, authentication, and serverless Edge Functions that route queries to AI. Utilizing pgvector within PostgreSQL, the team implemented a RAG (Retrieval-Augmented Generation) pipeline using the client's verified knowledge base to prevent AI hallucinations. Key features include an interactive AI chatbot, smart question suggestions, bilingual support, newsletter integration, and a privacy-first architecture.

Hallucinations eliminated by architecture. Importing the client's verified knowledge base into a RAG pipeline made answer accuracy a structural property of the system rather than a prompt-engineering hope — the single hardest requirement in AI-assisted health education. 100% user anonymity, designed in from the discovery phase: no accounts, no identity trail, no barrier between a teenager and a reliable answer. The "first question" barrier solved in UI. The smart suggestion system converts unaskable questions into one-tap starting points — a small feature doing the heaviest lifting in the product. Bilingual (PL/EN) from launch, with an architecture that makes further languages a content task, not a rebuild. A growth loop that respects anonymity — opt-in newsletter capture alongside a zero-identity core experience. Serverless, single-database architecture on Supabase: no infrastructure to manage, scalable by default, ready for global expansion and continuous knowledge-base growth. Proof that a sensitive-domain AI product can be built responsibly at startup speed — grounded retrieval, privacy-by-design and expert-curated content.

Primary: Generation Z — teenagers and young adults navigating puberty, body changes, relationships and intimate health, who have questions they will not ask a parent, a teacher or a friend. They are digital-native, privacy-conscious, allergic to being lectured, and currently getting their answers from search engines and social media. ZeroShame meets them where they already are — in a chat window — with the one thing those channels can't offer: verified content and zero exposure. Secondary: parents and guardians looking for a safe, age-appropriate resource they can point their child toward; teachers, school counsellors and youth workers who need a trustworthy supplement to a curriculum that is often incomplete, politically constrained or absent entirely; NGOs and public-health organisations working on adolescent health literacy. Market context: with 74% of young people reporting difficulty accessing sex education, this is not a niche — it's an entire generation underserved by institutions, currently defaulting to the least reliable source available.

We solved the hardest problem in applied AI: not being wrong. In sexual health education, a plausible-sounding wrong answer is a safety incident. We designed the RAG pipeline and knowledge-base import so accuracy is enforced.