Nearfolk
Community resource information is scattered across government and 211 sources, written in bureaucratic language, and searching it often means handing a query to a US-based AI service.
Nearfolk is a free, privacy-first directory of community resources across Ontario. The architecture is the real story: public data from 211 Canada and open government sources, cleaned and made searchable, with bureaucratic language translated into plain English by an open-source model running on local servers in Kingston. Searches never leave that hardware.
That resolves a real procurement blocker for nonprofit, municipal, healthcare, and legal clients who need AI features but can’t accept a US-hosted service touching their data. I’ve already built and shipped this. Very few Canadian consultants can say that.
What it does
- Indexes public data from 211 Canada and open government sources
- Translates bureaucratic language into plain English using an open-source model
- Runs that model on local servers in Kingston, so searches never leave the hardware
- No data selling, no paywalls, no forced accounts, no tracking or analytics that identify users
What it demonstrates
- Data sovereignty backed by an actual shipped project
- Local AI inference at production scale
- A real, live product with real users