Claude Duet
Anthropic’s proactive personal agent: it works in connected apps on a cloud computer and returns briefs, files and approval requests. Launched as an invitation-only US Pro and Max research preview in October 2026.
- Developer
- Anthropic
- Introduced
- 2026-10-21all 2026 milestones
- Access
- US Pro and Max accounts; waitlist invitation
- Preview model
- Claude Opus 5.5
- Usage
- Shares Claude plan session and weekly limits; background work counts
Unlike a chat that waits for the next prompt, Duet can prepare morning briefs and continue delegated tasks after the app closes. Tasks shows work needing input; Make holds finished files. The user marks completion.
Anthropic says sending messages, purchases, signing documents and deletion require approval. Securities trading, bank-account transfers and accessing another person’s account remain off-limits even with approval. These are documented safeguards, not proof that errors cannot occur.
The getting-started guide, checked October 22, 2026, excludes Team, Enterprise and Microsoft 365 from the preview. Disconnecting an app stops future access but does not erase memory of previous conversations. This is Anthropic’s product, not the unrelated open-source coding project with the same name.
Connected
- AnthropicAmerican artificial intelligence corporation
- Claudelarge language model family developed by Anthropic
- Prompt injectionAn attempt to make an AI system follow instructions embedded in untrusted material such as email, documents or websites instead of the authorized user’s task.
- AI data retentionThe rules governing which AI inputs, outputs, logs and derived records remain in a system, for what purpose, and for how long. Not using data for training does not mean that it is not stored.
- AI agents for business automationAgents that take actions across tools, from filing tickets to placing orders.
- Claude CodeAnthropic's agentic coding tool that reads a codebase, edits files, runs commands and verifies its work from the terminal, IDE or web.