AI Mental Health Chatbot

The premise was simple: what if someone who couldn't access therapy, whether due to cost, availability, or stigma, could still get real psychoeducation and evidence-based coping tools through a conversation?

This wasn't about replacing therapists. It was about filling the gap for the majority of people who will never see one.

Built using the Gemini API, the chatbot is designed around psychoeducation rather than diagnosis or treatment. It can walk users through grounding techniques, explain cognitive distortions in plain language, help someone identify stress triggers through guided conversation, and point toward appropriate resources when the situation warrants it.

The hardest design challenge was safety. Any AI operating in a mental health context has to be extremely careful about what it says and what it doesn't say. I spent significant time building guardrails and thinking through edge cases, particularly around crisis scenarios, before the core functionality.

The project deepened my interest in the intersection of AI and healthcare, and directly informed the MindSet app design work that followed.

---

Outcome: Built a working AI mental health chatbot using the Gemini API. It's designed for psychoeducation and evidence-based coping tools, with safety guardrails for crisis scenarios.

What I took from it:

• In sensitive domains, safety design isn't a feature you add at the end. It's the foundation everything else is built on: I spent more time on guardrails than on core functionality.
• The mental health gap isn't about quality of care. It's about access. Most people who need support will never see a therapist, and that's the problem worth solving.
• Building this made me realise how much the framing matters: psychoeducation feels different to the user than diagnosis, even when the underlying information is the same.