A field note · ~5 min read
Rewind.ai didn't fail. That's almost the more interesting story. It raised real money, shipped a real product, captured years of real users' screens — and then quietly pivoted toward Limitless. The original app still runs. It just isn't where the founders' attention is anymore. For a tool whose entire pitch was "we'll remember everything for you, for years," that's almost the same as shutting down.
Rewind launched as a Mac app that recorded your screen continuously, made it searchable, and let you ask questions across months of work. The hook was compounding value: every day made the database more useful than the last.
Then the same team launched Limitless — a pendant, a phone-call recorder, a different product entirely. Investment, talent, and roadmap attention shifted. The Mac app didn't get killed. It got something arguably worse: it became the legacy version of an idea the company had moved past.
Most "memory app shutdown" anxiety focuses on the wrong scenario. Outright shutdowns are rare and usually well- telegraphed. The far more common ending is the pivot — founders find a bigger market, raise on the new thesis, and the product you depended on stops getting the updates it needs to keep working as the OS, the AI providers, and the rest of the stack moves around it.
You don't get a refund for years of compounding value. You get a slowly degrading version of a product that used to feel essential.
Short answer
Rewind didn't die, it pivoted — and for a years-of-capture tool that's effectively the same thing. The lesson isn't "avoid memory apps". It's: keep capture and memory separate, prefer tools that export to plain text, and store the distilled meaning in a vault that doesn't depend on any one provider staying in love with its original product.
A vault that imports from the tools you already use — and exports clean if it has to. So no one company's roadmap can take your memory with it.
Meet Konshus