Most people using Claude for work still think of it as two separate tools: a chat window for quick questions, and Cowork for the bigger, file-heavy jobs. As of September 16, 2026, that split no longer exists. Anthropic folded Claude Cowork and Claude chat into a single interface, and most practitioners have not yet noticed what that actually unlocks.
If you are still opening a new Cowork session every time a task grows past "quick answer," you are working the old way. Here is what changed, and the exact setup to use it properly.
What Actually Changed When Claude Merged Chat and Cowork
Claude Cowork and Claude chat now live in one window instead of two separate modes you had to choose between upfront. A single conversation can start as a quick question and, without switching apps or losing context, expand into a multi-step task that reads your files, uses your connected apps, and produces a finished document.
The rollout started on Pro and Max plans and is expanding to more plans over the following weeks, according to Anthropic's own announcement and coverage from TechCrunch and Fortune. Two new capabilities shipped alongside the merge: Claude Docs and Claude Slides, which let you ask for a document or a presentation directly inside a normal conversation instead of opening a separate app.
---
Claude Chat vs Claude Cowork: When to Use Which Mode
The modes still exist, they are just no longer separate front doors. Chat is still the right choice when one person needs one fast answer, no files, no follow-through. Cowork is the right choice the moment a task needs your files, your connected apps, or more than a single back-and-forth to finish.
A useful rule of thumb from practitioners already using the merged interface: if you would have opened a second browser tab to check a file or a spreadsheet while chatting with Claude, that is your signal to switch to Cowork instead of copy-pasting content back and forth.
---
Three signals that a task belongs in Cowork, not Chat:
--- It touches more than one file or a folder you have connected
--- It needs a connector (Gmail, Google Drive, Slack, Microsoft 365)
--- The output is a document, spreadsheet, or slide deck you will reuse
How to Set Up Claude Docs and Slides Without an IT Team
Setting up the unified workflow takes about ten minutes and needs no API keys, no code, and no IT ticket. Download the Claude desktop app, sign in with a paid plan, and open Settings then Connectors to toggle on the apps your work already lives in.
Connectors turn on with a single toggle: Google Drive, Gmail, Slack, and Microsoft 365 are the ones most marketers, ops managers, and freelancers reach for first. Once a connector is on, Claude can read from it and write back to it inside the same conversation, with the permission mode you choose controlling how often Claude asks before it acts.
There are three permission modes worth understanding before your first real task. The most cautious mode has Claude check in before almost every action, which is the right default while you are still learning what the merged workflow can do. A middle mode lets Claude batch low-risk steps together and only pause for anything that sends a message, publishes content, or touches money. The most autonomous mode lets Claude work through an entire multi-step task with minimal check-ins, which is useful once you trust a specific recurring workflow but is not where you should start.
Switching modes costs nothing and takes one click from the mode selector in the chat box, so there is no reason to commit to one setting permanently. A freelancer drafting client proposals might run cautious mode for anything involving a client's actual data, then switch to the more autonomous mode for pure drafting work that never leaves a private folder.
---
Here is a complete prompt you can paste as-is to test the new Docs feature on a real task:
Try This Prompt:
"I need a one-page client update document. Pull the latest campaign numbers from [connected Google Sheet name], summarise what changed since last week in plain language, flag anything that needs the client's decision, and format it as a document I can download as a PDF. Keep it under 400 words and write in a direct, no-fluff tone."
---
Turning a Chat Thread Into a Cowork Task Mid-Conversation
The single biggest workflow change practitioners report is that a task no longer has to start in the right mode. You can ask a quick question in chat, realise halfway through that you actually need Claude to touch three files and build a slide deck, and keep going in the same thread instead of restarting in a new Cowork session and re-explaining everything.
This matters most for anyone who has ever lost twenty minutes re-pasting context from an old chat into a new Cowork window. That workflow tax is gone. The context you built up asking clarifying questions carries straight through into the file-and-connector work.
Here is what that looks like on a real task. A content marketer starts by asking Claude to check whether last week's blog post has been indexed yet, a simple chat question. Claude answers, and the marketer follows up by asking it to also pull this month's traffic numbers from the connected analytics dashboard and turn the whole thing into a one-page performance summary for a Friday stand-up. Under the old split interface, that second request meant abandoning the chat and starting fresh in Cowork, re-explaining which post, which date range, and which format. In the merged interface, Claude already has all of that context and simply keeps going.
The practical effect is that practitioners stop pre-planning which mode a task belongs in. You just start talking, and the tool scope expands to match what the task actually needs.
What This Means for Solo Practitioners and Small Teams
Freelancers and one-person marketing functions get the biggest relative gain from this change, because they never had a dedicated ops or IT person managing tool sprawl in the first place. Before the merge, a solo content creator juggling five clients often avoided Cowork entirely because remembering which mode had which client's connectors felt like more overhead than it saved.
With connectors and permission modes living inside the same conversation you already use for quick questions, that overhead mostly disappears. A single Project per client, holding that client's brand voice, recurring instructions, and connected folder, replaces what used to require manually re-establishing context every session.
For small teams of three to eight people, the bigger unlock is that recurring workflows, like a weekly content calendar review or a monthly client reporting cycle, can now be built once as a documented process inside a shared Project, rather than living in one person's head or one person's chat history that nobody else can see.
Common Mistakes When Switching to the Unified Workflow
The most common mistake is connecting every app at once "just in case." Claude only asks permission for connectors you have actually turned on, so an unused connector is pure risk with no upside. Turn on only what a specific task needs.
The second mistake is skipping permission mode settings. There are modes that control how often Claude checks in before acting, from asking before every step to working more autonomously. Teams handling client data should keep the more cautious mode on for anything touching email or shared drives, and loosen it only for low-stakes drafting work.
The third mistake is treating a Project like a folder, the old mental model. A Project now works more like a standing conversation that holds your brand voice, recurring instructions, and shared context, so recurring work like campaign reporting or content review should live inside one Project rather than being rebuilt from scratch in every new thread.
Try It Now: Your First 20-Minute Cowork Task
Pick one recurring task you currently do by hand every week, something like a weekly content performance summary or a client status update. Connect the one app that task depends on, start a normal chat, describe the finished outcome you want in one sentence, and let Claude propose its approach before it starts.
Review the proposed plan once, approve it, and watch where Claude actually saves you time versus where you still need to step in. That gap is exactly where your next automation opportunity is hiding.
The Takeaway
The merge is not a redesign for its own sake. It removes the tax of deciding upfront which mode a task belongs in, and that tax was real: every time you guessed wrong, you lost the context you had already built. Setting this up properly today, before your whole team figures it out independently, is a small edge that compounds every week you use it.
We understand AI. We understand you better. With UD by your side, AI doesn't feel cold.
Reviewed by the UD AI team.
Know Your Own AI Skill Level?
You just learned how to set up a faster Claude workflow. The next question is whether your prompting and workflow habits are actually keeping up with what the tools can now do, or whether you are still leaving speed on the table. UD's free AI IQ Test scores exactly that in a few minutes, and we'll walk you through every step of turning the result into a concrete improvement plan for your workflow.
