Claude now picks Chat or Cowork for you

1

Chat and Cowork are becoming one Claude.

Deciding whether a task belonged in Chat, Cowork, or a Project has been one of the most confusing parts of using Claude, and Anthropic is removing that choice. In the new version, Claude can move from a quick question into a larger task involving files, research, reports, or spreadsheets in the same conversation; that could mean going from a deposition transcript to a draft outline with page-and-line references to spot-check, or from a case-file question to a chronology, without starting over.

The change is rolling out in stages, beginning with Pro and Max plans, while Team plans have it in beta and Enterprise plans get it once an administrator turns it on. If you still see separate Chat and Cowork options, your account hasn't switched yet, and once it does, there is no switching back.

Mine has already switched, and what I like most is that I no longer have to think about whether something belongs in a project. When I ask for a task or an output that needs Cowork, Claude turns the conversation into a Cowork task on its own.

2

Opus 5.5 and Sonnet 5.5 are the current models.

Opus is built for complex work that calls for careful judgment, and Sonnet for well-scoped everyday work, including documents, slides, and spreadsheets. In my own use, Opus 5.5 is a huge improvement in writing; sentence structure, word choice, and length all feel more natural and much easier to follow.

For my daily work, my default is Opus 5.5 at Medium effort (the effort setting sits in the model menu next to the send button), and I recommend the same for you and your team; it's just that good. The only time I move off Medium right now is to build my website, which I do with Opus 5.5 at High. Monitor your usage, and if using Opus 5.5 as your default is consuming your usage too quickly, move some work to Sonnet 5.5.

The model menu next to the send button, with Opus 5.5 selected and Effort set to Medium.

3

Documents, decks, and designs open in a sidebar beside any conversation.

You can ask Claude to build the work product, then refine a specific section by keeping the conversation going or by editing it directly yourself in the sidebar, which brings the work we did on formatted documents, tracked revisions, evidence tables, timelines, and client-facing explanations into one place instead of a separate tool.

I built this newsletter in a document in the sidebar, and being able to give Claude edit directions and watch them land in the document right away has been a huge time saver.

Docs, Slides, and Design are out of beta and available on every plan, including Free.

A document open in the sidebar, beside the conversation that's building it.

4  What I'm using

Claude Duet is a proactive agent, and I've been using it myself.

Anthropic opened a research preview of Duet on October 21, and so far I'm very happy with it and genuinely impressed. What I love about Duet so far is how much it keeps track of: my to-do list, follow-ups on emails, whether responses have come in, action items, and the next steps on a workflow. It's extremely comprehensive. If you've seen xAI's Grok Bot, Meta's Muse, or ChatGPT's Dots, Duet is Anthropic's entry in that same category: an agent that takes ongoing responsibility for your tasks and keeps working between conversations.

Duet runs on its own computer in the cloud, so it keeps working while your laptop is closed. It sends a brief each morning, checks all your connected apps every few hours, drafts replies and next steps for your approval, and takes on larger tasks you hand it by message, such as setting up meetings or building a deck, then tells you when it's done or needs a decision. It also brings over your memory, skills, and connected apps from your Claude account, and Anthropic says what it learns from early users will shape how Duet and the main Claude app work together, which is a development I'm excited to watch.

Access is limited for now. Duet is open by waitlist only to Pro and Max subscribers in the United States (you can join the waitlist here), and Team and Enterprise plans aren't included yet, so many firm accounts can't use it. Duet shares your plan's usage limits with the rest of Claude, its background checks count toward those limits, and larger tasks take more turns and more background work. Once you start using Duet, check your usage over the first few days to see whether the rate it draws down your limits is one you're comfortable with.

Before it sends, pays, signs, or deletes anything, it shows you the action and waits for your OK, and a few actions stay off-limits even with your approval, including moving money between accounts and trading securities. Anthropic also cautions that Duet "may misread a thread or miscalibrate what's urgent." Because Duet reads your inbox, it also reads your client emails, so I'd treat connecting it the way you'd treat giving any new vendor access to client files. If you'd like help thinking that through before you join the waitlist, let me know.

5

Dashboards and Motion are new, depending on your plan.

Anthropic added both on October 8, and both are in beta. Dashboards turns a plain-language question into a live dashboard connected to your data, such as Salesforce, that stays current as the data changes, and you can click any number to see the query behind it; it's available on paid plans. For a small plaintiffs' firm, I'd build two of these right away: one that tracks cases by stage, and one for revenue and expenses.

Motion turns a prompt into a short animated explainer you can edit or download as an MP4, built from your own text, charts, and images, and it's available only on Team and Enterprise plans. On Enterprise plans, an administrator has to turn both on.

6

Two features worth knowing if you missed them

Neither of these is new since the workshops, but both are easy to miss.

Longer cloud tasks keep running after you leave the page. Transcript review, source comparison, and file organization are good candidates. Claude can keep working on eligible cloud tasks after you close the page or your laptop; if a task needs files or applications on your own computer, Claude Desktop has to stay open.

Memory is easier to see and manage. Under Settings, Claude lists what it remembers as individual topics you can edit or delete, which makes the About Me, working-style, and voice-profile work from Day One easier to keep current as your practice changes. Memory still only reduces repeated setup; the case-specific files, instructions, and sources you want Claude to rely on belong in the project itself. I've looked through my own memory list, and while I haven't needed to edit anything yet, it's a useful way to see how much Claude is capturing about my workflows, tasks, conversations, and projects.

The rule from the workshop has not changed

Claude drafts; you decide.

As Claude takes on more of the work, checking quotations, authorities, page-and-line references, formatting, computed dates, and case status before anything leaves your desk matters more, because there is more of it to check.

Work with me

If you want help applying any of this to your own practice, I offer one-on-one coaching over Zoom, and Shemia and I would be thrilled to bring an in-person training built for your team. Book a free 30-minute call below, or reply with a few times that work and I'll set one up. If the workshop was useful to you, please share it with colleagues and firms that could benefit.