OPENAI CALLED ITS NEW AI AGENTS “DOTS” — AND THEY CAN WORK ACROSS THOUSANDS OF APPS..
OpenAI is no longer positioning AI as something you open, ask a question and close. With Dots, the company wants an agent that keeps working after you leave.
THIS IS NOT ANOTHER CHATBOT
OpenAI introduced Dots on September 29, 2026, at its annual DevDay conference.
The basic idea is surprisingly different from ordinary ChatGPT.
You give a Dot a goal.
Then you leave.
The Dot can continue working in the background, using its own cloud computer, browser and connected apps. OpenAI describes Dots as proactive assistants that can keep working across complex projects and everyday tasks.
The change is subtle but important:
Chatbot → waits for you
Agent → works for you
That is the business transition OpenAI is trying to accelerate.
ONE DOT CAN CONNECT TO MORE THAN 4,000 APPS
OpenAI says Dots can connect to more than 4,000 apps through its app ecosystem.
That means the agent does not need to become a replacement for every piece of software you already use.
Instead, it can sit above them.
Your email can remain your email.
Your project-management software can remain your project-management software.
Slack can remain Slack.
Microsoft Teams can remain Teams.
The Dot becomes the layer connecting them.
That creates a much bigger opportunity than building another standalone productivity application.
THE REAL PRODUCT IS THE CONNECTION BETWEEN APPS
Imagine a product launch.
Normally, information is scattered across:
Email
↓
Slack
↓
Google Drive
↓
Project-management software
↓
Spreadsheets
↓
Customer feedback
↓
Calendar
A human has to move between all of them.
A Dot can potentially move through those systems itself.
It can collect information, update work, prepare documents and return to the human only when a decision or approval is needed.
The interface is no longer the individual application.
The interface becomes the agent.
OPENAI GAVE IT ITS OWN COMPUTER
This is one of the most important technical details.
Each Dot gets its own cloud computer and browser.
The user can open that computer and inspect what the Dot is doing. Dots can also be given permission to connect to the user's own computer when necessary.
That makes the architecture fundamentally different from a normal chatbot.
A chatbot produces an answer.
An agent needs somewhere to do things.
So OpenAI is effectively giving the AI:
brain → model
hands → computer/browser
memory/context → connected apps
instructions → user rules
That is much closer to a digital worker.
THE DOT CAN KEEP WORKING WHILE YOU DO SOMETHING ELSE
OpenAI says a Dot can take a project and continue working on it while handling other projects.
It can also send progress updates, ask questions and request decisions when necessary.
For example, OpenAI describes a developer whose Dot monitors customer feedback, identifies recurring requests, works on smaller fixes, tests them and prepares pull requests for review.
The human does not have to supervise every individual step.
Instead:
Goal → agent works → agent reports → human reviews
That is a completely different productivity model.
OPENAI WANTS THE DOT TO LEARN HOW YOU WORK
Dots are designed to become familiar with a user's goals and standards through interaction and feedback.
OpenAI gives another example involving a product launch.
If the scope of the launch changes, the Dot can work through the implications, revise launch materials and update related documents rather than waiting for the user to manually tell it what to change everywhere.
The economic idea is important.
The AI is not simply remembering facts about you.
It is learning how work should be done.
That makes the relationship potentially much stickier.
THIS COULD MAKE SWITCHING AI AGENTS HARDER
Suppose you spend six months teaching a Dot:
your writing style
your preferred reporting format
your customers
your projects
your approval rules
your internal terminology
your workflows
your standards
At that point, switching to another AI agent becomes more difficult.
Not because your files are locked inside OpenAI.
But because the agent has accumulated operational context.
That creates a potentially powerful form of product retention.
Dots CAN ALSO WORK THROUGH SLACK AND TEAMS
OpenAI says users can communicate with their Dots through ChatGPT on desktop, web and mobile.
They can also message them through Slack and Microsoft Teams, with text messaging planned for the future.
That matters because the agent does not need to live inside one application.
You could start a project in ChatGPT.
Discuss it with your team in Slack.
Then receive an update from the same Dot later.
OpenAI says the Dot carries context across these channels.
THE BUSINESS IMPLICATION IS HUGE
Think about how software has historically worked.
A company buys:
CRM
ERP
email
project management
analytics
customer support
HR software
accounting software
Each application has its own interface.
Humans are effectively the integration layer.
The human takes information from one system and puts it into another.
AI agents attack that inefficiency.
The new architecture could become:
Apps → Agent → Human
rather than:
App → Human → App → Human → App
The human moves up one level.
OPENAI IS TRYING TO BECOME THE ORCHESTRATION LAYER
This is probably the most important business angle behind Dots.
OpenAI does not need to replace every application.
It needs to become the system that coordinates them.
If the Dot can work across thousands of applications, OpenAI can potentially sit above an enormous existing software ecosystem.
That gives it a different strategic position.
Instead of competing with every productivity application directly, it can make those applications usable through one intelligent layer.
SPECIALIST DOTS GO EVEN FURTHER
OpenAI is also previewing specialist Dots for organizations.
These are not generic personal assistants.
Companies can give a specialist Dot its own identity, credentials and access to specific systems.
The Dot can then be assigned a defined responsibility.
OpenAI says it has already been testing specialist Dots internally for areas including:
procurement
invoice processing
email marketing
customer support
commercial contracting
That is essentially the beginning of an AI workforce architecture.
THE AGENT COULD BECOME A COMPANY EMPLOYEE WITHOUT BEING AN EMPLOYEE
Imagine a company with:
Sales Dot
Finance Dot
Customer Support Dot
Marketing Dot
IT Dot
Each one has access to different systems.
Each one has different rules.
Each one reports to humans.
The company could potentially create specialized digital workers without hiring a separate person for every repetitive workflow.
OpenAI says specialist Dots are initially being introduced through focused enterprise pilots.
MICROSOFT IS ALREADY PART OF THE STRATEGY
OpenAI says it is working with Microsoft to integrate specialist Dots with Microsoft Agent 365, Microsoft's enterprise governance and security controls.
The goal is to let companies manage Dots using tools they already have.
This is strategically significant.
Enterprise adoption is rarely just about whether an AI model is intelligent.
Companies need:
permissions
identity
security
auditability
governance
access controls
If OpenAI can plug Dots into those existing systems, deploying agents becomes easier for large companies.
OPENAI IS NOT GIVING THE DOT UNLIMITED POWER
There is an important limitation.
OpenAI says users can create Custom Rules that specify what a Dot may do independently, what requires approval and what is blocked.
Sensitive actions can require explicit permission.
OpenAI specifically says actions such as changing a password remain with the user.
This matters because the more autonomous an agent becomes, the more expensive a mistake can be.
A chatbot giving a bad answer is one problem.
An agent sending the wrong email, deleting data or changing an account is another.
PROACTIVE RESEARCH HAS ITS OWN RESTRICTIONS
When a Dot is working in the background without the user actively directing it, OpenAI says its proactive research can use connected apps in read-only mode.
That means it can inspect information but cannot simply start changing everything.
OpenAI says users can monitor background work through an Activity View and redirect the agent when necessary.
The architecture is therefore:
autonomy + boundaries
rather than unrestricted autonomy.
THIS IS ALSO AN ATTEMPT TO MAKE AI STICKIER
ChatGPT has traditionally been something people open when they need it.
Dots introduce another relationship:
AI that is already working before you return.
That changes the perceived value of a subscription.
You are no longer paying only for access to a model.
You are paying for ongoing work.
That is a much easier concept to connect to professional productivity.
THE SUBSCRIPTION MODEL COULD CHANGE
OpenAI says Dots are initially rolling out to Pro and Business Premium users in eligible markets, with Enterprise, Edu and Healthcare users able to try a beta when their workspace administrator enables it.
The first Dot is included in those plans at no additional cost.
OpenAI also says that, in the future, users will be able to add more Dots and scale their output by increasing either speed or the amount of work performed per month.
That creates an obvious future monetization structure:
one agent → included
more agents → paid
faster agents → paid
more work → paid
The agent becomes a consumption unit.
THIS IS WHERE THE “DOT” NAME MAKES BUSINESS SENSE
A Dot is not supposed to be one giant AI that does everything.
It can eventually become one of many.
OpenAI explicitly says it envisions teams of Dots working together on behalf of users.
That could produce an unusual software model.
Instead of buying more software seats:
one human + multiple AI agents
A company might eventually have dozens or hundreds of specialized agents operating under a much smaller number of human supervisors.
OPENAI IS ENTERING THE AGENT RACE DIRECTLY
The timing is important.
Meta launched its own always-available AI agent, Muse, earlier in September.
Google and other companies are also pursuing agents capable of interacting with applications.
OpenAI's Dots therefore arrive as part of a much broader shift from AI that generates content to AI that executes tasks.
The competition is no longer simply:
Who has the smartest chatbot?
It is increasingly:
Who can build the most useful digital worker?
4,000 APPS IS MORE IMPORTANT THAN 4,000 FEATURES
There is a subtle distinction here.
A normal software company might compete by adding more features.
OpenAI is instead trying to expand the number of places its agent can operate.
Every additional integration increases the number of tasks a Dot can potentially complete.
That creates a network effect around integrations:
More apps → more possible tasks → more useful agents → more users → more demand for integrations
The value of the Dot therefore grows with the ecosystem around it.
THE HUMAN BECOMES THE APPROVAL LAYER
This may ultimately be the biggest change.
Today:
Human does work → AI helps
Tomorrow:
AI does work → human approves
That reverses the traditional productivity relationship.
The human is no longer necessarily the person executing every step.
They become the person setting objectives, reviewing important decisions and handling exceptions.
The agent handles the operational middle.
MAACAT PERSPECTIVE
OpenAI's Dots are interesting because the product is not really the AI model.
The model is the engine.
The real product is the ability to give an agent a goal and let it operate across the software you already use.
4,000+ apps → one agent → continuous work → human approval when necessary.
That creates a new layer above existing software.
Email does not disappear.
Slack does not disappear.
Project-management tools do not disappear.
The human simply stops being the person manually moving information between them.
And if OpenAI eventually lets companies deploy teams of specialist Dots, the economic model changes again.
Software has traditionally been sold as:
one application → many human users.
The agent era could increasingly look like:
one human → many AI workers.
That is why Dots may be more important to OpenAI's business than another improvement in chatbot quality.
ChatGPT answered questions.
Dots are being built to carry the work.
