In partnership with
You don't have to decide where a task belongs in Claude any more. On Wednesday Anthropic folded Cowork into the normal chat app. One Claude. You describe the job, it picks the tool.
Anthropic's own reason: people used both, and "the frustrating part was deciding where a task belonged". I've lived in Cowork since it came out and I'd agree, the mode switch was the tax. Quick question, chat. Real work with files, Cowork. Plenty of mornings you started in the wrong one.
Pro and Max first, rolling out over the next few weeks on web, desktop and mobile. Team and Free after. If your app still shows two modes, it hasn't reached you yet.
What carries over
Everything, per Anthropic: chats, projects, artifacts, connectors, skills, files. Recents now shows conversations and tasks in one list. Global Instructions moved into Settings under Instructions, so if you had a house rule in there, go and check it survived. Two things didn't make it: adding from GitHub, and search doesn't find old Cowork tasks yet.
Background work now runs in the cloud, so you can close the laptop and it keeps going. Anything that touches your actual desktop still needs the desktop app open.
The toggle to check first
There's a new permission setting and it defaults to Manual: Claude asks before it acts. Auto lets it keep working and only check in when something needs a closer look.
Leave it on Manual for a week. Same rule as a new hire on day one: it earns access by doing the job in front of you. Switch to Auto per task (per task, not everywhere at once) once you've watched it do that task properly three times. I think most people will flip it to Auto on day one because the asking is annoying, and that's exactly the habit that bites.
Now a quick word from today's sponsor.
1,000+ Proven ChatGPT Prompts That Help You Work 10X Faster
ChatGPT is insanely powerful.
But most people waste 90% of its potential by using it like Google.
These 1,000+ proven ChatGPT prompts fix that and help you work 10X faster.
Sign up for Superhuman AI and get:
1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals
Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning
Right. Back to it.
The doc and the deck, without leaving Claude
Two new things you can ask for in any conversation. Claude Docs is a living document: you write it with Claude and edit it in place, then share it by link or export it. Claude Slides drafts a deck, you edit it, present from Claude or download it as PowerPoint or PDF. Design is in the conversation too now, for mockups.
The first job I'd run, because it uses all of it at once:
Take the folder [name] and the last 20 emails with [client]. Build me a one-page project update as a Claude Doc: what's done, what's blocked, what's due this week, with the source next to each line. Then turn the same content into a five-slide deck. Ask me before you send or share anything.That used to be a Cowork job, and then a chat job, and then you'd go open a slides tool. Now it's one message.
Don't burn your usage by lunch
Long tasks eat more of your usage than a chat message does. Hand it a two-hour job on a Monday morning on Pro and you'll feel it by lunch. Batch the heavy ones.
Make it stick
Move your standing rules into Settings, Instructions, now that Global Instructions live there: how you write, what it never sends, which folder is the source of truth. One place, every conversation, and you stop repeating yourself.
That's one door instead of two, for one task. An owner has the whole business's admin to hand over, and now it all goes through the same door.
If you own a business and want the boring work off your plate, that's what Run On Claude is for. Every repeat job you'd hire for, set up on Claude so it runs on autopilot, with the install files. You keep the decisions. Owners only. $147 a month until member 50.
Quick one.
Has the merge reached you yet, and did anything break? Hit reply.
Talk soon,
Zephyr
