Overview

FeelWell tries to make mental-health support feel inviting instead of clinical. It blends a voice-enabled chatbot, browser games, real global mental-health statistics, and bite-sized education on nutrition and exercise into one platform — built on Python/Flask and React/Next.js, hosted on Azure with PostgreSQL. The bet: people engage with wellness when it's interactive, not when it's a wall of text.

Tech Stack

Challenges

• Building a voice-enabled chatbot that's responsive and accessible in the browser.
• Sourcing and presenting trustworthy global mental-health statistics clearly.
• Folding mood-lifting games into a web app without it feeling gimmicky.
• Making education on nutrition and exercise genuinely engaging.

Solution

P5.speech handles voice recognition and Rivescript drives the chatbot logic, so users can talk to a conversational assistant, explore visualizations of mental-health trends, and play browser games like Block Rider. Flask powers the APIs, React/Next.js the experience, with Azure hosting and PostgreSQL behind it.

Outcome

FeelWell pulled education, play, and supportive conversation into one accessible place, and the interactive mix did its job — people stayed and explored rather than bouncing. It made the case that "engagement" and "wellness" don't have to pull in opposite directions.

What I'd do differently

Rivescript made the chatbot predictable but rigid — it can't handle anything off-script. I'd replace it with a grounded LLM kept inside safe, wellness-appropriate guardrails, and add real analytics to measure whether the games and content actually shift how people feel; right now "increased awareness" is a claim I couldn't yet put a number on.