This is a submission for the World's Largest Hackathon Writing Challenge: Building with Bolt.
You’ve probably come across the love languages quiz.
Maybe even the attachment styles one.
I wanted to build something just as approachable, but focused on helping people discover the type of therapy that might best suit their needs and preferences.
MindFLTR: Building a Therapy Match Tool with Bolt (My First Full-Stack Build to Deployment Project)
When I joined the World's Largest Hackathon, I had a simple idea: create something that helps people explore therapy in a way that feels straightforward, accessible and informed. I wanted to test an idea I’d been sitting on, something to help people understand what kind of therapy might suit them best, because therapy is not a one-size-fits-all.
That idea became MindFLTR — a short quiz that gives users a sense of which types of therapy might suit them best, based on how they think, feel, and approach various scenarios in life. It's not meant to replace professional advice, but it can be a useful place to start when seeking support for mental health and wellbeing.
What the App Does
MindFLTR is a mobile-first web app. Users take a short multiple-choice quiz, and the app calculates how well their responses align with seven recognised therapy types. These include CBT, Psychodynamic, Humanistic, DBT, Solution-Focused Brief Therapy, EMDR, and Expressive Arts Therapy.
After the quiz, users get:
- A ranked breakdown of therapy matches, shown as percentages
- A personalised summary explaining their top match, generated using OpenAI
- A downloadable report
- The option to sign up by email for future updates
- Accessibility features: font size adjustments, high contrast mode, keyboard navigation
It’s built to be calm, clear, and genuinely helpful, especially for people who’ve felt lost trying to figure out where to start with therapy.
I intentionally made the quiz logic, weighting and scoring neuro-inclusive, so that neurodiversity is accounted for.
MindFLTR is meant to spark reflection, offer clarity, and hopefully give people a better sense of where to start when it comes to therapy.
Building with Bolt
I used Bolt.new to build the app from the ground up. The frontend was built with React and Tailwind, and Supabase handled backend tasks like email collection and Edge Functions.
Here’s what I implemented:
A modular component structure for the landing page, quiz flow, results, and email screen
A custom scoring system that turns user answers into normalised match percentages
Supabase Edge Function to call the OpenAI API, sending user traits and their top therapy match to generate a personalised summary. Getting that API integration working smoothly and securely was a big part of the build.
State management across the app to handle navigation and accessibility settings
A floating toolbar for font and contrast adjustments
What Informed the Design
I didn’t want to build just another quiz. Before I started, I used Perplexity to synthesise peer-reviewed research around therapy-seeking behaviour, common misconceptions about modalities, and how digital tools often fall short.
That early research shaped the logic, tone, and structure of the quiz. My background in psychology (from undergrad through to my current PhD) definitely influenced the way I approached the background research, matching system and how I framed the results. The goal was to inform and encourage, not overwhelm.
Therapy isn’t one-size-fits-all. It took building this tool to realise just how hard it is to design for nuance — and just how worth it that challenge is.
What I Learned (and Had to Figure Out)
This was my first time wiring up both frontend and backend. Getting Supabase and Bolt to talk to each other, handling Edge Functions, and working with environment variables — all new and challenging.
The scoring logic felt simple at first, but it wasn’t. I wanted it to feel balanced. It wasn’t just a matter of tallying points. I wanted the output to be meaningful, not just cosmetic or a surface-level match, so I spent time refining how percentages were calculated and presented.
For the AI summaries, I spent time refining the prompts to guide the AI toward more semantic clarity and emotional nuance.
Accessibility features took careful planning — especially dynamic font scaling and full keyboard support.
The hardest part wasn’t building — it was knowing when to stop.
There’s always more to add, but the time constraints of the hackathon forced me to draw a line between what's essential and what's just nice to have. I learned to scope responsibly, make tough calls, and trust that simplicity can still be impactful.
Why Bolt Helped
Bolt gave me the space to focus on what I wanted to build, rather than worrying about setup and structure. AI-generated components were surprisingly helpful for getting past the initial blank canvas. Real-time previews and smooth Supabase integration made the workflow easier than I expected.
Using Entri
I also used one of the hackathon partners — Entri — to purchase a custom domain for the project: mindfltr.me. I bought it right at the end of the hackathon window, so I didn’t have time to connect it properly to the deployed version of the app.
Still, it felt like an important step, giving the project a real name and a permanent home. I plan to point the domain to the live version of MindFLTR as I continue developing it.
Looking Ahead
MindFLTR is just getting started. I’d like to build on it, improve the AI responses, add more therapy modalities (and adaptations), languages, and pay further attention to cultural sensitivity. I also have plans to include next-step guidance, and perhaps collaborate with organisations for this part. But the core of it is live, and it works.
I want MindFLTR to be more than just a quiz; I want it to be a starting point for exploring the full spectrum of support that exists for our mental health and wellbeing. There’s a whole universe beyond traditional talk therapy, and I think people deserve to know what’s out there.
Eventually, I’d love to expand MindFLTR to include lesser-known, emerging, and even immersive approaches — from nature-based practices to body-led modalities to future-forward therapies like VR-assisted exposure or guided simulations. As those fields grow, I want MindFLTR to grow with them; offering people a broader, more inclusive way to find the kind of support that fits.
For now, I’m proud of what it’s become. And I’m grateful to the Bolt hackathon team and all of their sponsors and partners for giving me the fuel to power forward and build the things I’ve been dreaming about.
MindFLTR demo
By Sola Janet Browne
Top comments (0)