In the long arc of human labor, tools have always extended what a person can do in a given hour — but they have waited, inert, for the hand to pick them up again. OpenAI's announcement of Dots at DevDay 2026 marks a quieter but more consequential shift: AI agents that continue working through the night, across thousands of applications, while their human counterparts attend to other things entirely. The question being posed to enterprises is not merely one of productivity, but of trust — how much of the ongoing machinery of work a person or organization is willing to place in the custody of a
OpenAI Launches Dots: Always-On AI Agents That Work Alongside Human Teams
An agent that remembers preferences and develops continuity over months may become harder to replace
Mark
So Dots are just ChatGPT that keeps running in the background?
Mimi
Not quite. ChatGPT waits for you to ask it something. A Dot keeps working on projects you've assigned it—monitoring for changes, updating documents, testing code—all while you're doing something else.
Luke
But what does "working" actually mean here? Can it modify files on its own, or does it need approval?
Mimi
It depends on what you've set up. For read-only work—checking for bugs, reviewing feedback—it can do that autonomously. For anything that changes something, there's an approval system that checks the action against your rules before it proceeds.
Mark
And if it makes a mistake?
Mimi
OpenAI says you should review consequential work. The agent can still get things wrong.
Luke
That's the real governance question, isn't it? A chatbot's mistake is a bad paragraph. An agent's mistake could alter records or send information to the wrong place.
Mimi
Exactly. That's why OpenAI built Custom Rules and Activity View into the product from the start, not as an afterthought.
Mark
What about the team collaboration part—ChatGPT Space?
Mimi
It's a shared workspace where employees and agents work on the same documents together. A Page can pull information from Slack, email, calendars and update automatically as things change.
Luke
So if I'm on a team and my Dot is updating a shared Page, what information about me becomes visible to my coworkers?
Mimi
Your private ChatGPT conversations stay private. But if the Dot writes something from your memory onto the shared Page, your coworkers can see it.
Luke
That's a significant privacy boundary to manage. Does OpenAI address that?
Mimi
They acknowledge it in their documentation. They warn that material someone shares could potentially affect another collaborator's memory or be eligible for training under that collaborator's settings.
Mark
Is this actually going to change how people work?
Mimi
If it works as designed, yes. Instead of executing every step yourself, you'd supervise a portfolio of work performed partly by agents. But that requires trust, governance and clear boundaries.
The Pulse
- The fundamental tension is not technical but existential: Dots do not wait to be asked, they keep moving, and that continuity challenges every assumption enterprises have built around human-initiated AI interactions.
- The disruption ripples outward — from individual workflows to team collaboration, as ChatGPT Space dissolves the boundary between what one person's AI knows and what the whole team can build on together.
- OpenAI is threading a careful needle on autonomy, restricting background activity to read-only access and requiring human approval before agents send messages, modify content or take consequential action.
- Specialist Dots — assigned organizational identities, credentials and defined business responsibilities — signal that enterprises may soon need to onboard, govern and eventually offboard AI agents the way they manage human employees.
- The competitive landscape is crystallizing fast: Meta's Muse crossed millions of downloads within weeks of launch, and a cohort of startups is converging on the same persistent-agent model, making this less a feature race and more a battle for long-term trust.
In the long arc of human labor, tools have always extended what a person can do in a given hour — but they have waited, inert, for the hand to pick them up again. OpenAI's announcement of Dots at DevDay 2026 marks a quieter but more consequential shift: AI agents that continue working through the night, across thousands of applications, while their human counterparts attend to other things entirely. The question being posed to enterprises is not merely one of productivity, but of trust — how much of the ongoing machinery of work a person or organization is willing to place in the custody of a persistent digital mind.
OpenAI introduced Dots at DevDay 2026 — a new class of AI agent built not around answering questions but around owning work over time. Where ChatGPT waits for the next prompt, a Dot monitors projects, uses software, responds to changing conditions and returns completed work for human review, all while its user is occupied elsewhere.
Dots run on GPT-6 Astra from their own cloud computers and browsers, connecting to more than 4,000 applications through OpenAI's plugin ecosystem. They communicate through ChatGPT, Slack and Microsoft Teams, and gradually learn a user's preferences and working standards. Sam Altman signaled they will eventually be reachable through additional messaging apps and by phone.
Paired with Dots is ChatGPT Space, a shared collaborative environment where employees, ChatGPT, Codex and Dots can work against the same context. Space replaces ChatGPT's existing Library for Pro, Business and Enterprise users and hosts Pages — documents that stay synchronized with connected tools as underlying information changes. The effect is to move AI work out of private chat threads and into a shared layer where teams and agents build on each other's progress.
OpenAI's use cases reveal its enterprise ambitions clearly. A developer-facing Dot can monitor customer feedback, scope and build fixes, and return pull requests with demonstration videos. A sales Dot can analyze prospect requirements, construct proofs of concept and keep proposals current as deals evolve. Content teams can hand a Dot an interview transcript and receive clips, show notes and social drafts, with editorial changes carried across all associated materials automatically.
Governance is built into the product's architecture rather than bolted on afterward. Background research is restricted to read-only access — Dots cannot send messages, alter content or control a user's machine without explicit permission. Custom Rules let organizations define which actions are permitted, which require approval and which are prohibited outright. An Activity View lets users inspect what a Dot has been doing, and an auto-review layer checks consequential actions against user instructions and OpenAI's safety requirements before proceeding.
The more ambitious layer is specialist Dots — agents assigned their own organizational identities, credentials and defined business responsibilities. OpenAI has already piloted agents in procurement, invoice processing, customer support and commercial contracting, and is working with Microsoft to integrate specialist agents into Microsoft Agent 365. If the model takes hold, enterprises may come to treat agents less like software tools and more like machine identities requiring onboarding, access policies and eventual offboarding.
OpenAI is not moving into open territory. Meta's Muse, launched weeks earlier, had already reached millions of downloads and topped app-store charts in the United States. Startups including Instinct, Manus and SpaceXAI are pursuing the same persistent-agent interface. The competitive question is no longer which model produces the best answer — it is which agent a person or organization trusts enough to hand continuous access to the machinery of their working life.
OpenAI announced Dots at DevDay 2026, a new category of AI agent designed to keep working long after an employee closes the chat window. Unlike ChatGPT, which waits for the next question, a Dot monitors projects, uses software, responds to changing information and brings completed work back for human approval—all while its human counterpart is doing something else entirely.
Dots run on OpenAI's GPT-6 Astra model and operate from their own cloud computer and browser. They can connect through OpenAI's plugin ecosystem to more than 4,000 applications and communicate with users through ChatGPT, Slack and Microsoft Teams. Over time, they learn an individual's preferences, standards and working habits. The agents will be accessible in ChatGPT on mobile and web, and Sam Altman said during his keynote that they will soon be available through other messaging apps and even as audio models over the phone.
But Dots are not simply individual assistants. OpenAI is pairing them with ChatGPT Space, a new collaborative layer where employees, ChatGPT, Codex and Dots can work against the same shared context. Space replaces ChatGPT's existing Library for Pro, Business and Enterprise users and acts as a home for Pages, files, presentations, spreadsheets and other team materials. Rather than keeping AI work confined to one person's private chat, teams can bring people and agents into shared documents, tag a Dot or ChatGPT for help, and continue building on work already completed by colleagues or other agents. Pages can remain synchronized with connected tools, letting teams create documents that update as underlying information changes.
OpenAI's examples make clear where it wants enterprises to deploy the technology first. For developers, a Dot can watch incoming customer feedback, identify recurring requests or bugs, scope smaller fixes, build and test them, and return completed pull requests accompanied by videos demonstrating the changes. For product and marketing teams, the agent can learn an organization's audience, positioning and creative standards, then revise launch material when the underlying product changes. A sales-focused Dot can compare an enterprise prospect's requirements against product documentation and account history, identify remaining technical tests, construct a proof of concept, suggest an appropriate solutions engineer and keep the proposal up to date as the deal changes. Content teams get a similar treatment: a Dot can ingest an interview transcript, find potential clips, prepare show notes, draft social posts and carry subsequent editorial changes across the associated materials.
The distinction matters for enterprises. Much of the first wave of generative AI adoption has required employees to repeatedly initiate interactions: open an assistant, supply context, request an output, review it, then start again when the situation changes. Dots are designed around continuity instead. They can keep several projects moving simultaneously, accept new work without forcing users into separate conversational threads and operate while their human counterpart is doing something else. The common thread is not generation, but workflow ownership. An employee does not merely ask the AI to produce one artifact. The employee establishes an objective and standards, while the agent monitors the underlying work and responds as circumstances change.
Governance sits at the center of the product rather than as an enterprise feature added later. When a user is not actively working with the agent, a Dot can inspect information in already connected applications and look for things that may require attention. OpenAI restricts that background activity: the tools used for proactive research are read-only and cannot send messages, modify content or control the user's computer or browser. Once actual changes or external actions enter the picture, permissions become more complicated. Organizations and users select which applications a Dot can access through ChatGPT's existing app controls. OpenAI is introducing Custom Rules, which can permit particular actions, require approval or prohibit them. An Activity View lets users inspect background work and intervene. An additional auto-review system checks potentially consequential actions against the user's instructions, Custom Rules and OpenAI's built-in safety requirements before determining whether the work can proceed autonomously or requires approval. Some sensitive operations, including changing passwords, remain reserved for the human user. Each Dot operates inside its own cloud computer, separated from the employee's machine unless the user explicitly connects the two.
The more ambitious enterprise play comes from specialist Dots. A personal Dot acts for one user. Specialist Dots instead receive their own organizational identity, credentials and access to company systems and are assigned a defined business responsibility. OpenAI says it has already experimented internally with agents working in procurement, invoice processing, email marketing, customer support and commercial contracting. It is beginning focused pilots with enterprises in which OpenAI engineers work directly with customers to define each agent's responsibilities, permitted tools and approval processes. OpenAI is also working with Microsoft to integrate those specialist agents into Microsoft Agent 365, allowing businesses to govern them through security and management systems they may already use. If that model works, enterprises may eventually treat agents less like software licenses assigned to humans and more like new machine identities that themselves require onboarding, credentials, access policies, monitoring and offboarding.
OpenAI is not entering an empty market. Over the past several weeks, a cluster of companies has begun converging on a remarkably similar idea: an AI assistant with a persistent identity, its own computing environment, access to a user's accounts and applications, memory of what matters to that person, and enough autonomy to act without continuous prompting. Meta's Muse, launched September 8, had already surpassed between 2.3 and 4.3 million downloads by late September, less than three weeks after launch, depending on which analytics provider's estimates you trust. Muse reached the top of both Apple's and Google's U.S. app-store rankings during this surge. Meta is already trying to carry that momentum into work, expanding Muse with small-business integrations spanning tools including Shopify, QuickBooks, Stripe and Canva. Startups including Instinct, Manus and SpaceXAI are chasing the same interface. Together these launches suggest that persistent personal agents are becoming a distinct product category—and potentially the next major interface battle in AI. The competitive question is shifting from whose model gives the best answer to which agent a person or business is willing to trust with continuous access to the machinery of everyday life and work.
Notable Quotes
Rather than keeping AI work confined to an employee's private chat, teams can bring people and agents into shared documents, tag a Dot or ChatGPT for help, and continue building on work already completed by colleagues or other agents.— OpenAI's description of ChatGPT Space functionality
A mistake made by a chatbot may produce an inaccurate paragraph. A mistake made by an agent operating authenticated business software can potentially alter records, send information or trigger downstream workflows.— OpenAI's governance documentation
Want the full story?
Read the original at VentureBeat ↗
Technology