Fossick

This started with my partner's parents. They were being sued, and I watched themspend night after night digging through their own computer for the documents their case depended on — contracts, letters, records they knew were somewhere on the machine but couldn't find. They remembered what each file was about; they just couldn't remember what it was named or which folder it was in. Filename search was useless, and uploading years of private legal documents to some cloud tool was out of the question.

That's where Fossick came from.

Fossick is a desktop app (Mac + Windows) that runs semantic search over your own documents, entirely on-device. You point it at a folder; it reads and embeds everything locally (all-MiniLM-L6-v2 + sqlite-vec) so you can search by what a file means, not what it's named. Nothing is ever uploaded — you can pull your Wi-Fi and it still works. It's search, not chat: no LLM, no cloud account for your files. It also OCRs scanned PDFs and images, which matters when half of what you're hunting for is a scanned contract.

Where I'm at: built solo, free during beta (no card), and I use it every day.

The honest challenge: I've got a product I believe in and basically zero distribution. Building it was the easy part; getting it to the people who need it — often at the worst moment of their lives, like my partner's parents — is the real work. Next milestone is my first 10 paying users. If you've taken a local-first / privacy tool from "it works" to "people paying,"

I'd love to hear what actually moved the needle. And if you try it, tell me what it fails to find — that's the feedback I want most.