Your browser can tell you where you were on a Tuesday afternoon. Ask it what you read there and, by default, all it has is the title.
Manny's take is blunt: browser history is broken, and the new AI fixes don't repair it, because every one of them asks you to decide ahead of time what you'll want to remember.
Memory that needs advance notice isn't memory.
Title. URL. 90 days.
Here's what the default actually keeps. Chrome's built-in history stores the page title and the URL, and its search matches only those. Chrome Help is just as plain about time: your History lists the pages you've visited "in the last 90 days." Older entries go automatically.
So the whole design fits in one line:
default history = title + URL + 90 days
That's a fine tool for one job, a log of where you went. Scrolling it for something you read feels like walking a graveyard: every stone has a name and an address, and nothing about what was there.
And you don't remember names and addresses. You remember the chart, the one sentence, the argument that changed your mind. Everyone has lost a page they needed. It's annoying every single time, and it is NOT a sign you're careless. The tool was built for a different job.
Conditions
"Hey Manny, this is old news. Chrome has AI history search now."
Fair, and credit where it's due. Chrome's "History search, powered by AI" lets you search in everyday language, and when it's on, Chrome stores the page contents on your device, encrypted, next to the title and URL. That's a real step. Microsoft is moving too: Copilot Journeys in Edge turns your recent browsing into topic cards.
Now read the conditions. Chrome's version is opt-in, works only if you're located in the US and use Chrome in English, and covers only sites you visit after you switch it on (Chrome Help); Edge's Journeys is turned off by default, starts from URLs and page titles, keeps a short summary per page rather than the full content, and deletes older Journeys, underlying data included, after 14 days (Microsoft Support).
Why so many conditions? Hard to know from the outside. His guess: what you read says far more about you than where you went, so the vendors tiptoe. That's a reasonable instinct, and privacy deserves the caution.
But look at what the conditions share. To get the memory, you have to turn it on before you need it, and if you didn't, the words on the page you're hunting for today were never kept.
Same habit as bookmarking. New settings page.
Read now, decide never
The clearest description of the problem doesn't come from a browser vendor. It comes from a 2022 Ask HN thread asking for "a fully searchable full-text history forever with little fuss," from someone who didn't want to save or annotate an article first. One line in it says everything: "My brain... does not know ahead of time what could be useful."
Nobody's does. That's how reading works. You usually find out which page mattered weeks later, when a colleague asks or a decision comes back around.
So the fix is boring, and boring is right: keep what you read, all of it, by default, for good, and let search do the work later. That's the job Datarag takes on. Datarag is a browser extension that saves and indexes everything you read online, so you can search your full browsing history in plain language — or ask an AI about it. Instead of a graveyard of addresses, you get the pages back, searchable by what they said.
You read it once. That should be enough.