Carevo

Inspiration

Walking into a nonprofit service event, you see the chaos immediately: volunteers with clipboards, paper forms stacked on folding tables, clients waiting while staff scramble to find last week's intake sheet. Data gets lost. People get missed. We kept asking ourselves: why are the communities that need the most help being served by the least modern tools? That question became Carevo

What it does

Carevo is an AI-powered case management platform built for nonprofits running service events. Volunteers can register clients in seconds by snapping a photo of a paper form or dictating information by voice — Claude AI handles the extraction automatically. The interface is multilingual (English, Spanish, and French), so language is never a barrier. Coordinators get a live dashboard where they can ask natural language questions like "how many Spanish-speaking families did we serve this month?" and get instant answers, no SQL required. Every service entry is tracked, searchable, and exportable as a PDF report.

How we built it

We built Carevo as a three-layer system. The frontend is a Next.js 16 app with Tailwind CSS and Supabase Auth, handling client registration, event management, and analytics. The AI service is a Python FastAPI backend that uses Claude's vision API to parse photos of paper forms and OpenAI embeddings to power semantic search across service notes. Supabase handles the database, authentication, and row-level security so each organisation only sees their own data. The whole frontend is deployed on Vercel, with the AI service running as a separate backend.

Challenges we ran into

Extracting structured data from messy, real-world paper forms took significant prompt engineering. We also ran into Next.js build issues on Vercel around static pre-rendering with dynamic hooks like useSearchParams, which required wrapping components in Suspense boundaries. Designing a role system flexible enough for super admins, nonprofit admins, and frontline staff, without making it complicated for volunteers to just show up and use it, was a real UX challenge.

Accomplishments that we're proud of

We're proud that a volunteer with no technical background can walk up, open the app, scan a form, and have a client registered in under 30 seconds. The multilingual support feels genuinely useful, not just a checkbox. And the natural language analytics dashboard asking a question in plain English and getting a real answer about your community felt like a real breakthrough moment for the team.

What we learned

We also learned that building for nonprofits means designing for people who are tired, understaffed, and doing the work because they care, so every second of friction you remove matters actually.

What's next for Carevo

We want to expand the AI analytics to support trend detection and proactive alerts, for example, flagging when a client hasn't been seen in several months. We'd also like to add offline support for events in areas with poor connectivity, a client-facing QR check-in flow, and deeper integrations with existing nonprofit CRMs. Most importantly, we want to put it in the hands of real organisations and let their feedback shape where it goes next.

Built With

• next.js
• python
• supabase
• vercel