Claude Duet: What Anthropic’s Proactive Agent Means for Brands

Model: Claude · Duet

On 21 October 2026, Anthropic opened a research preview of Claude Duet, a new app where Claude works as a proactive agent, for people on Pro and Max plans. Here’s what an assistant that can act before it is asked could mean for the answers people receive and for the brands those agents choose to name.

Source: Claude release notes — Claude Duet research preview for Pro and Max plans (21 October 2026)

What Anthropic launched

On 21 October 2026, Anthropic opened a research preview of Claude Duet: a new app in which, in its own words, Claude works as a proactive agent. The preview is available to people on the Pro and Max plans, and Anthropic has pointed to a companion “Introducing Claude Duet” post for the detail. The framing is deliberately modest — a research preview is an early look, not a polished product — so the honest read today is the direction of travel rather than a fixed set of features.

Still, the word that matters is “proactive.” Most AI tools wait for a prompt and reply. An agent described as proactive is one that can take the first step — surfacing a suggestion, starting a task, or flagging something worth your attention before you ask. It is the same agentic arc we have tracked across the field, from GPT-6 Pro’s parallel agents to the Claude plugin directory, now pointed at everyday assistance rather than one-off research runs.

What a proactive agent does

A proactive agent is one that does not only respond — it initiates. Where a chat assistant answers the question in front of it, a proactive agent can watch the context of your work, anticipate the next step and offer to take it. In practice that can look like a nudge, a draft prepared in advance or a shortlist of options put forward for you to pick from, rather than a blank box waiting for instructions.

Anthropic has said little beyond the “proactive agent” framing, so we will not pretend to a feature list. What matters for brands is the behavioural shift the phrase implies: an assistant that raises options unprompted is an assistant that, at times, names providers, products and sources before anyone asks it to. That is the same underlying judgement we examined in how LLMs decide which brands to cite — now potentially applied one step earlier in the conversation.

Who can use Claude Duet

It is worth being precise about availability, because it shapes how much changes today. Claude Duet is a research preview for people on the Pro and Max plans — Anthropic’s paid tiers — not a free default and not a general release. That means the early audience is committed, paying users rather than the broad public, and that the experience is expected to change as the preview develops.

We are deliberately not claiming this touches Google’s AI Overviews, AI Mode or any web-search surface — it is a new app within Claude, not an answer engine wired into search. What it does signal is where assistants are heading: from a tool you open and query toward one that works alongside you and speaks up on its own, much as we noted when a cheaper frontier Claude and a faster Claude Haiku 5.5 put capable models in front of more people.

Why a proactive agent changes discovery

When an assistant only answers, discovery happens at the moment a person asks “which tool should I use?” When an assistant acts first, discovery can happen earlier — the agent spots a need and offers a name before the question is even formed. That moves some brand exposure upstream, into moments the person did not explicitly start, which raises the stakes on being a name the model is willing to put forward unprompted.

It also means a single screenshot is a weaker read than ever. A proactive agent can take different routes depending on context, timing and what it judges useful, so what it suggests once may not be what it suggests next time. Measuring how Claude represents your category needs the same discipline we set out for interactive answers — watch the pattern across many runs, not one moment.

Where your brand can and cannot appear

There is no “rank in Claude Duet” to chase and no agent to submit your brand to. A proactive agent is a way of working, not a directory or an ad slot — you cannot buy a placement inside it. The place your brand can be named is where it always has been: in Claude’s ordinary output, drawn from the public sources and signals the model has learned to trust.

So treat a proactive agent as a more forward reader, not a new billboard. It does not add a surface to game; it changes when a name can come up. The brands that benefit are the ones whose facts a careful model can confirm quickly — the principle behind whether your brand is ready for AI search.

What this means for your content

The response is not to tune your site for one app but to make your expertise unmistakable to any version of Claude. Spell out who you serve, the problems you solve, your credentials and the evidence behind your claims, in clean, structured content a model can read across your own site, your profiles and third-party sources. An agent that acts proactively still leans on the same learned trust — it simply applies it sooner.

Consistency does the heavy lifting. When your key facts match everywhere a model looks, a careful model can name you without hedging — whether it is answering or offering. That is the same foundation we set out in building a GEO strategy, Google’s own llms.txt guidance, and our overview of how AI visibility tracking works.

Does a proactive agent change recommendations?

Not the rules of them. A proactive agent changes when and how a name can surface, not whom Claude trusts. When it names a provider, product or study, it is still drawing on the sources it treats as authoritative — earned through clear, corroborated, widely consistent evidence rather than a clever prompt. Acting first does not lower that bar; if anything, an agent putting a name forward unasked needs to be sure of it.

That is why the fundamentals matter more as agents improve, not less. The competition is still to be the source a careful model reaches for — the same logic we traced in how stronger models sharpen recommendations, when Gemini 4 Argon raised the ceiling, and when Perplexity went multi-model. A more forward agent raises the reward for real evidence; it does not create a shortcut around it.

How to check your visibility across AI agents

With proactive agents like Claude Duet beginning to roll out, it is worth taking a baseline now. Here is a short, practical way to see how today’s models handle the questions around your brand — today is 23 October 2026:

1. List your real questions. Write down the questions your customers actually ask before choosing you.
2. Ask Claude directly. Run those questions in Claude and record which brands, providers and sources it names.
3. Watch what it offers. Note where an assistant suggests options you did not explicitly ask for, and whether you are among them.
4. Note the sources. Where a model references pages, check whether your own content is among them.
5. Compare across models and regions. Repeat in ChatGPT, Gemini and Perplexity, and in your key markets, since answers differ.
6. Track the trend. One run is a snapshot; the value is in watching movement as a new agent beds in.

How reconnAI tracks AI answers

reconnAI monitors how the leading AI models — ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI Overview — answer questions across regions, and re-baselines that tracking whenever a platform ships something that changes behaviour. A proactive agent is exactly that kind of moment, because it changes when and how a brand can be named, not just how a question is answered.

It joins a busy run of shifts we have followed lately, from GPT-6 Pro’s parallel agents and ChatGPT’s new default to GPT-6.1 Sol and Gemini 4 Argon. If you want to see how Claude represents your brand when people ask about your category, get in touch with our team, explore how AI visibility tracking works, or read our overview of ChatGPT advertising for where paid placement is heading.

Claude Duet: your questions answered

What is Claude Duet? It is a new app from Anthropic, opened as a research preview on 21 October 2026, in which Claude works as a proactive agent — an assistant that can take the first step rather than only answering prompts.

Who can use it? The research preview is available to people on Claude’s Pro and Max plans. It is an early preview rather than a finished, general-release product, and the experience is expected to change.

What does “proactive agent” mean? It describes an assistant that can initiate — surfacing suggestions, starting tasks or flagging next steps — rather than waiting for every instruction. Anthropic has shared little beyond that framing so far.

Can I pay to have my brand suggested in Claude Duet? No. It is a way of working, not a placement. Claude names brands from the public sources it trusts, so there is nothing to submit or buy — the work is clear, consistent evidence.

How do I track my visibility across models? Baseline the questions your customers ask, run them across Claude, ChatGPT, Gemini and Perplexity, and watch how the answers change over time — which is what reconnAI automates.

About reconnAI

reconnAI tracks how the major AI models represent topics and sources across ChatGPT, Claude, Gemini, Perplexity, Copilot and Google AI Overview — across multiple regions. We monitor how those models answer and how they change over time, so you can stay ahead of shifts in the AI landscape.

Last updated: 23 October 2026.