Histiq

Histiq is a browser extension that gives you semantic search over your own browsing history and the PDFs you read, with every bit of processing happening locally on your device. Instead of remembering the exact title of a page you read three weeks ago, you describe what it was about and Histiq finds it by meaning. It runs on Chrome, and on Edge through its native support for Chrome extensions, and it needs no account of any kind to work. Installation comes from the Chrome Web Store and the trial starts the moment it's installed, so finding out whether it fits your reading habits is a two-minute decision.

The problem is one every heavy reader hits constantly. Browser history search only matches page titles and URLs, so the article you remember as the one about caching strategies for mobile apps is unfindable if those words never appeared in the title. You end up scrolling through days of history entries, or re-googling and hoping the same result surfaces, or admitting the page is simply gone from your life. Histiq attacks this by indexing what pages actually say, then letting you search the way you actually remember things, by concept rather than keyword. The gap between how memory works and how history search works is the entire product.

The technical approach is what makes it notable. Histiq ships an embedded machine learning model, Xenova MiniLM at roughly 25MB, that runs directly inside your browser. As you browse, it converts page content into 384-dimensional vectors and stores them in your browser's local IndexedDB. When you search, your query goes through the same conversion and matches are ranked by cosine similarity, all on your device. No external API, no model server, no cloud. The site even invites you to verify the claim yourself by inspecting the extension's network requests, since there are none for your history, which is a far stronger privacy statement than any policy page could make.

The indexing is more thorough than a naive scrape. It captures full page content including collapsed sections and text hidden behind read-more elements, which is exactly the content that vanishes when a lesser tool grabs only what's visible on load. It also handles the text layer of PDFs up to 50MB, pulling research papers, reports, and manuals into the same searchable pool as web pages. Everything indexes automatically in the background as you browse, so there's no save button to remember and no ritual to maintain. The one structural limit is that it can only know what you read after installing it, so the index grows more useful the longer it runs. The PDF support is the sleeper feature here, since papers and reports are exactly the documents people re-hunt most often and the ones ordinary history search handles worst.

It fits power users and researchers, anyone who reads heavily and later needs to relocate something half-remembered. Academics working through stacks of papers, developers who read documentation and blog posts all day, analysts, journalists, and students all fall into the natural audience. If you've ever spent twenty minutes trying to re-find a page you know you read, the pitch lands immediately. The local-first architecture also makes it viable for people whose browsing includes confidential material, legal research, medical reading, competitive analysis, since nothing about what you read ever leaves the machine.

That privacy posture is the real differentiator. Plenty of tools offer to remember everything you read, but most do it by shipping your reading history to their servers, which is exactly the data many people least want to share with anyone. Histiq's bet is that a small local model is good enough for personal-scale search, and it trades cloud horsepower for the guarantee that your history stays in your browser. For this particular job, finding things you personally already read, that trade looks sensible, since the index only ever needs to be as big as one person's reading life. A cloud service can search the whole internet, but it can't know which corner of it you personally read last Tuesday, and that's the only question Histiq is built to answer.

Pricing is trial-then-pay. Installing from the Chrome Web Store starts a 14-day free trial with no credit card required, after which it's $6.99 a month, $49.99 a year which works out to about $4.17 a month, or a $129 lifetime license with no recurring billing. There's no permanent free tier, but the lifetime option is a reasonable escape hatch from subscription fatigue for a tool designed to sit quietly in your browser for years. Contact goes through a form on the site.