Related stock research
Continue researching the companies
Connect this market insight with company earnings, business trends, risks, and institutional ownership.
OpenAI used its September 29 Developer Day to move the AI competition into a more aggressive phase.
The company introduced Dots, always-on AI agents that can pursue user goals across applications with limited supervision.
Dots are powered by GPT-6 Astra and can operate through tools such as Slack, Microsoft Teams, Codex and ChatGPT Work.
OpenAI says the agents can manage ongoing projects, prepare materials, monitor tasks and continue working in the background rather than waiting for a new prompt every time.
That makes Dots much closer to a digital worker than a conventional chatbot.
The launch puts OpenAI into direct competition with Meta’s Muse and intensifies pressure on enterprise software companies that have spent years selling workflow, collaboration and automation products.
What Makes Dots Different From ChatGPT
Traditional ChatGPT interactions are session-based.
A user asks a question.
The model responds.
The interaction ends unless the user continues.
Dots are designed to persist.
A user gives an agent a goal.
The agent can continue working across applications.
It can monitor progress.
It can perform research.
It can return when something needs human review.
That persistence is important because many business processes are not single tasks.
They involve follow-ups, changing information, multiple stakeholders and work that unfolds over days or weeks.
An always-on agent can potentially own part of that process.
Why Slack and Teams Integration Matters
Enterprise work already happens inside communication platforms.
Employees coordinate projects in Slack and Teams.
They share files.
They assign tasks.
They discuss decisions.
By placing Dots inside those environments, OpenAI can become part of the workflow without requiring users to leave the tools they already use.
That reduces adoption friction.
It also increases competitive pressure on Microsoft, Salesforce, ServiceNow, Atlassian and other software companies that want their own AI agents to become the orchestration layer for enterprise work.
The battle is shifting from “who has the best model” to “which agent becomes the default layer across business applications.”
OpenAI Is Using Its Existing Distribution
The company says Codex and ChatGPT Work already have more than 35 million weekly users.
ChatGPT itself has about 1.2 billion consumer users, according to Reuters.
That gives OpenAI enormous distribution.
A new enterprise product does not have to begin from zero.
Existing ChatGPT users can become Dots users.
Existing coding users can connect agents to development workflows.
That installed base is one reason Dots can become significant quickly even in a crowded enterprise market.
The Meta Muse Comparison
Meta’s Muse generated millions of downloads and showed that consumers are willing to try agents that complete real-world tasks.
Dots aims more directly at work.
The competitive difference is partly distribution.
Meta has Facebook, Instagram and WhatsApp.
OpenAI has ChatGPT, Codex and enterprise AI relationships.
Both companies are trying to turn an AI assistant into a persistent agent.
The next battle will be about trust, permissions, integrations and economics.
Capability alone will not decide the winner.
The Demo Glitches Matter
Reuters reported that live demonstrations of Dots experienced technical problems.
That is not unusual for early-stage software.
It is still important because autonomous agents face a higher reliability standard than ordinary chatbots.
A chatbot can give a bad answer.
An agent may change data, contact people or trigger workflows.
The cost of failure is therefore higher.
Enterprise customers will care about audit logs, permissioning, identity, rollback and monitoring as much as they care about benchmark performance.
Safety Is Becoming a Commercial Requirement
OpenAI says users must provide explicit approval for sensitive actions such as password changes.
The company also says business data is not used for training without consent.
Those controls are part of a broader shift in the AI industry.
As agents become more autonomous, safety is moving from research into product design.
Customers want to know not only what an agent can do, but what it is prevented from doing.
That creates opportunities for cybersecurity, identity and governance companies.
It also creates pressure on AI platforms to prove they can operate safely inside enterprise environments.
Why Enterprise Software Stocks Should Care
An always-on agent can overlap with functions that companies currently buy from multiple software vendors.
Project tracking.
Customer support.
Workflow automation.
Coding assistance.
Document creation.
Business intelligence.
Scheduling.
Internal search.
If one AI agent can coordinate across several systems, customers may reconsider how many point solutions they need.
That does not mean Salesforce or ServiceNow becomes obsolete.
Large enterprises still need systems of record, permissions and specialized workflows.
But the user interface may shift away from individual applications and toward agents.
That can change where software value accrues.
Microsoft Faces a Complicated Position
Microsoft is both a partner and a competitor.
OpenAI’s Dots can operate through Teams.
Microsoft is also building its own persistent agents through Copilot.
That creates a familiar AI-industry structure in which companies cooperate on infrastructure or distribution while competing for the customer interface.
For Microsoft investors, the risk is not that OpenAI disappears from the ecosystem.
It is that OpenAI captures more of the user relationship inside the Microsoft stack.
For OpenAI, Microsoft distribution remains valuable even as competition increases.
The Economics Are Still Unclear
Always-on agents can consume much more compute than chatbots.
They can run continuously.
They can monitor many data sources.
They can use tools and browsers.
They can generate long sequences of model calls.
That raises the question of pricing.
A flat subscription may not cover extremely heavy usage.
Usage-based pricing can control economics but makes costs less predictable for customers.
The winners in enterprise agents will need to make the value of automated work exceed the cost of continuous inference.
What to Watch Next
Watch Dots adoption among Pro, Business Premium and Enterprise customers.
Watch real-world reliability.
Watch permission and safety incidents.
Watch integrations beyond Slack and Teams.
Watch pricing and usage limits.
Watch how Microsoft, Salesforce and ServiceNow respond.
Watch Meta Muse adoption in business workflows.
And watch whether customers begin replacing point software with agent-led workflows.
OpenAI’s announcement is not simply another AI product launch.
It is a direct attempt to make the agent—not the application—the primary interface for work.
If that model succeeds, enterprise software will be forced to adapt around it.