Histiq
Local semantic search for browser history & PDFs
I kept losing pages I'd read. I'd remember what an article was about but not its title, and browser history only matches titles and URLs. So I built Histiq.
It indexes the full text of the pages and PDFs you read and lets you search them by meaning. Type a rough description and it finds the page even if none of those words were in the title.
The rule I set: it has to be 100% local. It runs a small quantized model (all-MiniLM-L6-v2, ~22MB) on-device via transformers.js and ONNX WASM. Nothing is uploaded. You can open DevTools and watch a search make no external calls.
Stack notes for anyone curious: keyword search first with a semantic fallback; text stored in IndexedDB; PDFs indexed by extracting the text layer in chunks; the model runs in an offscreen document to stay single-threaded and skip the cross-origin isolation requirement.
Honest status: launched, but installs are low and I'm looking for the first real users. It's paid (14-day free trial, no card, then $6.99/mo or $129 one-time). Not open source right now.
I'd love feedback, especially on search quality and pricing. What would make you trust a tool like this with your history?
Site: https://histiq.com
5 Comments
The local-first approach is probably the strongest part here. Browser history is one of those datasets where users may want intelligence, but not at the cost of giving someone else access.
The challenge is creating the moment where someone realizes “I need this” before they lose another important page.
I think you're right. That's the hardest part—people have to install it before they actually need it. It’s like insurance, which is always hard to sell.
What really helps is making it useful right away. Once you install it, it searches through your old browser history. So instead of thinking "maybe I'll need this someday," you immediately see that it found a page you lost last week.
I'm also realizing that losing a page isn't a disaster, but it happens all the time. For people who read a lot online, it’s just a small, constant annoyance that keeps repeating.
That's a fascinating shift. It's always interesting when users quietly overturn one of your strongest assumptions.
I'd be interested in comparing a few observations outside the thread. What's the best email to reach you on?
Sure, here is my email. You can write to me at support@histiq.com.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.