Dots vs Claude Duet vs Grok Bot: Three Work Agents
All 3 keep working after you close the app. Only 1 is built as a team of bots. How their computers, approvals and availability actually differ.
Last Updated: November 2026
Dots, Claude Duet and Grok Bot are three work agents that keep running on a computer after you close the app. That is what puts them in the same conversation. They are at very different stages, and that matters more than any feature list. One is documented and generally available, one has an inconsistent rollout, and one is a restricted research preview.
AiBuildrs builds custom AI systems for mid-market businesses. The team has completed over 200 successful AI implementations, producing $4.7M in documented client cost savings.
Grok Bot is a team of named bots on a shared cloud computer. OpenAI Dots and Anthropic Claude Duet are the other two work-agent products people now compare with it.
They are at very different stages, and that matters more than any feature list. One is generally available on paid plans, one has an inconsistent rollout, and one is a restricted research preview.
Key Takeaways
- Only Grok Bot is a team: Named bots run in parallel and share a workspace. The other two are single agents.
- Computers differ: Grok Bot shares one per account. Dots is reported to give each agent its own.
- Duet is a preview, not a product: US only, Pro and Max, by waitlist, with Team and Enterprise unsupported.
- Duet checks rather than runs: It looks at connected accounts every few hours instead of working continuously.
- All three ask before acting: Sending, paying, signing and deleting sit behind approval in each.
- Sourcing is uneven: Grok Bot has full documentation. Dots and Duet details come largely from coverage.
Those six points place the three products against each other. The sections below take each one in turn, compare their computers and approval models, and end with who might reasonably pick which.
What is Grok Bot in this set?
Grok Bot is xAI's team of always-on agents, introduced in beta (2026) in August 2026.
You create named bots that work in parallel rather than queueing. They keep going after you close the app and resume with your preferences, files and logins in place.
The documentation (2026) describes a persistent cloud computer with a browser, a filesystem and a terminal, shared by all of an account's bots. A shared workspace lets bots hand work between themselves.
The bots drive a real browser. That lets them work through sites and apps with no API at all, signing in and clicking through interfaces as a person would.
One caveat belongs with that design. Separate bots are workflow boundaries rather than security boundaries, since they share the one computer.
What is Dots?
Dots is OpenAI's agent product, announced at DevDay in September 2026. Agents take goals, learn preferences through feedback, and keep working in the background.
Reporting describes each dot getting its own cloud computer and browser, and connections to several thousand applications. The breadth of integration is the headline claim.
The important qualifier is that these details come from press coverage rather than from documentation AiBuildrs has been able to verify. Rollout and per-user limits have been reported inconsistently.
So treat the capability picture as provisional. Check OpenAI's own pages at the moment you need a decision rather than relying on a comparison written earlier. AiBuildrs re-checks all three before advising on any of them.
What is reasonably consistent across coverage is the shape: background agents, preference learning over time, and wide application connectivity.
What is Claude Duet?
Claude Duet is Anthropic's proactive agent, launched in October 2026 as a research preview. That status is the most important thing about it.
Availability is narrow. It is limited to the United States, for Pro and Max subscribers, by waitlist, and Team and Enterprise are not supported.
It also works differently from the other two. Rather than running continuously, it checks connected accounts every few hours and acts on what it finds.
Its approval model is cautious by design. It seeks approval before sending messages, making payments, signing documents or deleting information, and Anthropic says it does not directly see passwords or sign-in tokens.
Anyone presenting Duet as generally available is describing something that does not exist yet. State the preview status whenever you name it. AiBuildrs treats it as a product to watch rather than one to deploy.
How do computers, approvals and work style differ?
The computer is where the three genuinely diverge. Grok Bot puts all of an account's bots on one shared machine. Dots is reported to give each agent its own. Duet is not described as a persistent VM at all.
Work style follows from that. Grok Bot parallelises across bots. Dots runs continuously in the background. Duet works in periodic passes.
Approvals converge. All three stop before consequential actions, which suggests the industry has settled on that boundary rather than competing on autonomy.
Muse sits outside this set, being a personal agent rather than a work one, and the Grok Bot comparison covers it properly.
Who might pick each one?
Pick Grok Bot if the work splits into parallel streams and touches systems with no API. The team model and the browser are what you are buying.
Pick Dots if breadth of application connectivity is the binding constraint, and you can tolerate an uncertain rollout picture while you evaluate.
Do not pick Duet yet, in the sense of planning around it. A US-only research preview with no Team or Enterprise support is something to watch rather than deploy.
And weigh the sourcing difference seriously. Grok Bot can be evaluated against documentation. The other two largely cannot, which is a real consideration when you are committing a team's workflow.
Where none of them reaches into your own databases or internal tools, the constraint is architectural, and AiBuildrs builds bespoke AI systems for that case.
Frequently Asked Questions
Is Dots the OpenAI agent?
Yes. Dots is OpenAI's agent product, announced at DevDay in September 2026. It is built around agents that take goals and keep working in the background while learning your preferences. Coverage reports each agent receiving its own cloud computer and browser, with connections to several thousand applications. Those details come from press rather than documentation we could verify, so confirm them on OpenAI's own pages before relying on any of it.
Is Claude Duet the same as Claude chat?
No. Claude chat is Anthropic's conversational assistant. Claude Duet is a separate proactive agent that checks your connected accounts every few hours and acts on what it finds. Duet is also far more restricted. It launched in October 2026 as a research preview, limited to the United States, for Pro and Max subscribers by waitlist. Team and Enterprise are not supported at all.
Which of these is built as a team of bots?
Grok Bot. It is the only one of the three designed around multiple named agents working in parallel and handing work between themselves through a shared workspace. Dots and Duet are single-agent products, however capable each one is individually. If your reason for looking at agents is that work splits into streams, that structural difference matters more than any individual capability.
Do they all keep working offline?
They all keep working after you close the app, which is the property people usually mean. Grok Bot's bots are always-on and resume with your files and logins in place. Dots runs in the background continuously. Duet is the exception in pattern rather than principle. It checks connected accounts every few hours rather than working without pause between those checks, so there are gaps by design.
Which one has the most documentation?
Grok Bot, by a clear margin. xAI publishes product documentation covering how it works, its vault and secure sign-in, approvals and security, and skills and routines. Dots and Duet details come substantially from launch coverage and reporting. That gap matters when you are evaluating. Documented behaviour can be checked against a page, while reported behaviour has to be taken on trust from a third party.
Can any of them reach systems without an API?
Grok Bot can, because its bots drive a real browser on their cloud computer and can sign in and click through interfaces directly. Dots is reported to work similarly with its own browser. Duet's documented behaviour centres on checking connected accounts rather than general browser automation. Browser-driven work is powerful but fragile, since it breaks whenever a page layout changes.
Do any of them act without asking?
None of them is designed to, on anything consequential. All three gate actions such as sending, paying, deleting or signing behind a human decision. That convergence suggests the boundary is settled across the industry rather than being a point of competition. What differs is how much else they do freely between those approval points, which is worth testing on your own work.
Which is safest for business data?
That depends on what you are protecting against, and none has a simple answer. Grok Bot's documentation is explicit that its bots share one computer and are not security boundaries, which is useful honesty but a real constraint. Duet reports not seeing passwords or sign-in tokens directly. Dots has the least verifiable public detail, which is itself a consideration for a regulated environment.
Where does Muse fit in this comparison?
Outside it, mostly. Muse is Meta's personal agent for everyday tasks rather than a work-agent product. It is reached in its own app or in WhatsApp, and runs on a dedicated Secure VM per user. It belongs in a comparison about personal assistance rather than team workflow. The Grok Bot versus Muse article covers that contrast properly rather than squeezing it in here.
Should I wait before choosing?
For Duet, yes, since a research preview with no Team or Enterprise support is not something to build a workflow on. For the other two, a limited pilot is reasonable, starting with read-only tasks and keeping consequential actions behind approval. What you should not do is commit a team's process to a product whose rollout and limits are still being reported inconsistently.
Executive Summary
Grok Bot is the only one of the three built as a team. Named bots work in parallel and hand work to each other on one shared computer.
Dots is OpenAI's background agent product, reported to give each agent its own cloud computer and browser and to reach several thousand applications.
Claude Duet is the most restricted. It is a research preview, United States only, for Pro and Max subscribers by waitlist, and it checks connected accounts periodically rather than running continuously.
All three gate consequential actions behind approval. None of them is designed to send, pay or delete without asking.
What Should You Do Next?
Judge these three on stage of maturity before features. One is documented and available, one is still being reported, and one is a preview you cannot deploy.
If the work needs to reach into your own systems rather than public web applications, talk to AiBuildrs about what that actually requires.
People Also Read
- Grok Bot vs Muse: Work Teammates or a Personal Agent
- What Agentic AI Is, and How Grok Bot and Muse Use It
About the Author
Jerry Jariwalla is the founder of AiBuildrs and creator of the Growth Signal Intelligence framework. With over 22 years in digital marketing and multiple successful business exits, Jerry has spent the past decade leading AI implementation programs for mid-market businesses across professional services, recruitment, membership organizations, and traditional industries. AiBuildrs has completed over 200 successful AI implementations using a workflow-first methodology and is trusted by leaders at YPO, Vistage, Tiger 21, and C12 executive peer organizations.
Expertise: AI Strategy, AI Implementation, Workflow Automation, Custom AI Development, Voice AI, Offshore Engineering, B2B Sales Intelligence, Mid-Market AI Adoption
Connect: LinkedIn
Disclaimer: This content is for informational purposes only and does not constitute professional business or technology advice. ROI outcomes vary based on industry, existing systems, and implementation commitment. Contact AiBuildrs for a consultation regarding your specific situation.
Jerry Jariwalla
Jerry Jariwalla, Founder & CEO of AiBuildrs, builds custom AI infrastructure, voice AI, automation systems, and The AEO Engine. View Jerry Jariwalla's Full Profile →