What jumps out to me is less the feature itself and more the fact that it took this long to make memory consistent between chat and Cowork. That feels like the kind of thing users would assume already worked this way. If Claude is going to be pitched as a system that can follow you from a quick chat into a cloud task, then having separate memory islands would be a pretty awkward footgun.
The part I’d actually want to test is how well this behaves with real work context. “Less re-explaining” sounds great until memory starts surfacing stale assumptions, half-right project details, or old client preferences you no longer want carried forward. Anthropic saying sensitive topics are off by default is reassuring in theory, but in practice I’d still want to know how easy it is to audit, delete, and scope memory per task. That’s where these systems usually get messy.
There’s also a subtle product signal here. Unifying memory across web, desktop, and mobile makes Claude feel less like a chat app with a few extra modes and more like a persistent assistant. That’s the direction everyone in this space is pushing toward. The catch is that persistence only feels smart if the controls are obvious. Otherwise it turns into “Claude remembers things” as a marketing line and “Claude remembered the wrong thing” as a support problem.
I think the more interesting question is whether Anthropic can make memory useful without making it spooky. For developer workflows, that means predictable inheritance across contexts, clear opt-outs, and no surprises when a cloud task picks up prior chat history. If they get that right, it’s a real improvement. If they don’t, this just becomes another reminder that stateful assistants are much harder to get right than stateless chat.
Reference: Anthropic update unifies memory feature across Claude Cowork and chat - 9to5Mac
