Meta Muse vs Claude Duet vs Grok Bot: Who Wins Personal AI

The personal AI agent race is officially on. Meta says 10 million people have already set up Muse, Anthropic launched Claude Duet on October 21, and Grok Bot and OpenAI's Dots are fighting for the same spot on your phone. Our short answer after using them: Grok Bot is Jordan's favorite right now, Google is the player everyone is underestimating, and the real money is in enterprise agents, not consumer ones.

We spent the news segment of Episode 56 on this with Ivan Lee, co-founder and CEO of Datasaur, who worked on Siri at Apple before starting his company. Here is what we took away.

What just launched

In a few weeks the category went from empty to crowded. Here is the field as of late October 2026:

• Meta Muse launched in September. On Meta's earnings call, Mark Zuckerberg said 10 million people have already set up a Muse agent (report). It pulls context from the places you already talk: Instagram, Messenger and the rest of Meta's apps.
• Claude Duet is Anthropic's answer, announced October 21 as a research preview. It runs on Claude Opus 5.5, watches connected apps like email and calendar, sends a morning brief and asks before it takes consequential actions. It is waitlist-only for US Pro and Max subscribers for now (coverage).
• Grok Bot from xAI works through accounts you already have, and reviewers describe it handling errands end to end, like comparing prices and placing orders.
• OpenAI Dots is OpenAI's persistent agent. Jordan's read is that it has not caught on the way the others have.

Watch the personal agent debate from Episode 56 (starts at 17:18) · Episode 56 notes and transcript

Which one we actually use

We have been rolling these out to our own teammates over the last few weeks, and Sam said on the show that we have had real, working use cases already. Jordan has tried several, and Grok Bot is his favorite so far. He was quick to add that the ranking could flip next week, because the category is moving that fast.

His bigger point was about data. The agent that wins is the one already connected to your life, and that is why he thinks Google is being written off too early:

I think we're highly underestimating Google because it's already connected to all these data sources like your email in our case, you know. So we're on Google stack. Right? So for us, that's very helpful.

— Jordan Metzner, Episode 56

Microsoft has not shipped its version either. Jordan expects an updated Copilot that does this kind of work, which would make the race even more about who owns your inbox and calendar.

Why Muse broke out first

Ivan has been waiting for personal agents since his Siri days, and his read on Muse was the most useful thing we heard all week. Enterprise agents charged ahead this year, starting with coding agents, but consumers were left with nothing:

We've been promised these travel agents. We've been promised these things that, like, book weddings for you and do all this party planning and everything else, and they weren't really capable. And I think Muse kinda, like, figured out a UI, has all this context direct from you... From your Instagram, from your Messenger conversations, and it was able to kind of take the first pass at it.

— Ivan Lee, Datasaur, Episode 56

That matches what we see. The winning feature is not a smarter model. It is context the agent already has, plus an interface people already open every day.

Ivan is on the waitlist for Duet, and he pointed out that Anthropic is historically "more enterprise coded," so its consumer take will be worth watching. He also noticed there was no cute mascot at launch.

The hot take: agents fit work better than life

Ivan's closing argument is the one we keep coming back to:

I think agents are just universally a better fit for the enterprise than they are for the consumer world at this point.

— Ivan Lee, Episode 56

His reasoning: companies already have operational playbooks written to train the next human hire, so an agent has instructions to follow on day one. Consumers do not. You and I want different airports, different seats, different everything, and nobody has written that playbook down.

He also put a number on why this matters for cost. An average chatbot query is 200 to 2,000 tokens, while an average agentic task is about 3 million tokens. Agents that run in the background all day will eat compute in a way chat never did, which is part of why Duet's background activity counts toward your plan's usage limits.

The best personal agent is one you build

The most interesting use case in the episode was not any of the big launches. Ivan built his own: he wrote an MCP server for his fitness bracelet, connected Claude to it, and now Claude knows how he slept and gives him a training plan each day, including telling him to rest when he is short on sleep.

Jordan's reaction was that this opens the door to open hardware you wire up to whichever agent you like. That is the builder's version of this race, and it is the one we are most excited about. If you want to try it, start with the tools we use every week: Claude, Claude Code and Claude Cowork, and see how we compare them in ChatGPT vs Claude and Claude vs Gemini.

Who we think wins

• Short term: Meta, on distribution. Context straight from Instagram and Messenger is hard to beat.
• Watch closely: Google, because it already holds the email and calendar data an agent needs.
• Our daily driver: Grok Bot, for now.
• Long term: enterprise agents, built on company playbooks, will create more value than consumer agents for the next few years.

We will keep testing these as they ship. New episodes come out every week, with a live build, a founder interview and the AI news that matters.