Claude Cowork finally remembers what you told the app in chat
TechCrunch Sarah Perez ● Covered by 3 sources
Claude now shares memory between chat and Cowork. So it stops making you repeat yourself every time you switch tasks.
Based on reporting by TechCrunch, Sarah Perez — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic is tying Claude’s chat memory and its Cowork memory together, and that should kill one of the most tedious parts of using an assistant: repeating the same context over and over. If you’ve been talking through an idea in chat and then moved to Cowork to actually do something with it, Claude is supposed to carry the thread with it now.
That makes the product feel less like two separate tools stitched under the same name. Anthropic’s own example is pretty ordinary, which is exactly the point: drafting an update for a manager, or building a conference agenda, without having to restate the city, headcount, or speakers every time you switch modes. The usefulness is not flashy. It’s the absence of friction.
The other change is more interesting than it first sounds. Claude will now add topics to memory as the conversation happens, instead of waiting to summarize everything at the end. That means the shared memory between chat and Cowork should update faster, even while the chat is still going.
Anthropic is also giving users a window into what Claude has kept. People can read, edit, or delete memory entries on any topic, and the company says Claude won’t save personal or sensitive information by default, including health data, race, ethnicity, religious beliefs, politics, gender identity, and more. Users can turn on an option called “include sensitive topics in memory” if they want, and Claude will also warn them when something sensitive gets stored. Some things are off limits no matter what, including government-issued ID and Social Security numbers, criminal history, immigration status, and anything that breaks the company’s acceptable use policy. The feature is on by default for Free, Pro, and Max plans across web, desktop, and mobile, with iOS and Android users needing the latest app version.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI feature: less theater, more memory. The industry has spent ages selling assistants that sound smart while forgetting the previous sentence, which is a charming design flaw if your product is supposed to help with actual work. Giving users control over memory is the part that matters most, because trust dies fast once software starts remembering the wrong things for the wrong reasons.
Read more about this at: TechCrunch