Anthropic launched Claude Duet on 21 October 2026, an AI agent for professionals and small business owners. It is a research preview, with a waitlist for Pro, Max and Ultra subscribers in the US. Before anyone on your team connects it to real accounts, use this checklist to test it safely on low-risk tasks first.
What changed
This is a new, separate product from Anthropic, not an update to the existing Claude chat assistant.
That is how The Verge describes it, so expect everyday business tasks to be the focus.
A research preview means the product is early-stage and may change, have bugs, or behave unpredictably. Treat it as a test, not a finished tool.
Access is by waitlist, on Pro, Max and Ultra plans in the US. Team and Enterprise plans are not included yet.
Before you install it
Because this is a new agent still in research preview, check with whoever manages company app approvals or IT policy before anyone installs it on a work device.
Use a secondary or test account rather than your main business account for the first session, since the agent is new and its behavior with connected accounts has not been widely vetted yet.
Pick one small, low-stakes task (for example, drafting a social post or summarizing a document) so you can judge the output clearly instead of testing many things at once.
First session checks
Do these in order the first time you open Duet, whether on iOS, Android, or Claude on the web.
Research preview agents often ask for account or data access upfront. Note exactly what it is asking to connect to before you approve anything.
Something you can easily undo or check, like drafting a reply or outlining a document, lets you see how it works without risk.
Draft a short [email/social post] about [topic]. Do not send or publish it, just show me the draft.
Anthropic says it asks before sensitive operations like sending messages or paying. Good behavior looks like the agent pausing to show you a draft or plan and asking 'should I proceed' before sending, posting, or purchasing anything. If it acts without asking, stop and review its settings.
Read the full draft or result, not just the first line, and check for factual errors, wrong names, or anything you would not want sent under your business's name.
Once you have seen it behave predictably on a reversible task, try something like checking a calendar or reading an email, still watching for confirmation prompts before it takes action.
Ongoing precautions
Note the date, the account, and what access you granted so you can revoke it quickly if needed.
Because it is a research preview, Anthropic may change features or access levels. Recheck connected accounts and permissions periodically rather than assuming settings stay the same.
The goal is to avoid someone enabling account access on their own before anyone has reviewed how the agent behaves.
Want me to look at yours?
If you have gone through this and want a second pair of eyes, book a free call. We open the account together and look at what is actually happening. No deck, no pitch.
A new free sheet most days. Get the next one: follow @ihamza.jameel on Instagram · YouTube
More free sheets
- RETENTIONAI Chat Privacy ChecklistNew · Checklist · 21 checks · 18 Nov 2026
- FILTERSSearch Console New Filters ChecklistNew · Checklist · 14 checks · 16 Nov 2026
- HUBChatGPT Business Hub Setup ChecklistChecklist · 13 checks · 11 Nov 2026
- TRACKINGGoogle Ads Tracking Check: When Conversions DropChecklist · 21 checks · 11 Nov 2026
