Key Highlights
- OpenAI has introduced Dots, always-on AI agents designed to pursue user goals across apps with limited supervision.
- Dots can work with Slack, Teams, Codex, and ChatGPT Work to research, analyze data, prepare documents, and build software.
- OpenAI says users can set rules for the agents, while sensitive actions require explicit approval.
OpenAI is expanding its focus on autonomous AI for businesses with the launch of Dots, a new type of always-on agent designed to work toward user goals across different applications.
The company unveiled Dots at its annual developer event in San Francisco as it pushes further into the enterprise AI market. The agents are designed to handle evolving tasks with less day-to-day supervision, including projects where requirements change over time.
Dots Designed to Handle Ongoing Projects
According to OpenAI, Dots use its advanced GPT-6 Astra model and can manage projects as they develop. For example, an agent could update a sales proposal when customer requirements change or create a working demonstration for review.
Users can communicate with Dots through platforms such as Slack and Microsoft Teams. The agents can also use OpenAI’s Codex and ChatGPT Work tools to research information, analyze data, prepare documents, and build software.
The approach reflects the growing interest in AI systems that can move beyond answering individual prompts and instead carry out multi-step work on behalf of users.
“Growing quickly is crucial for OpenAI’s plans to spend tens of billions of dollars on AI data centers and keep pace with rivals such as Anthropic and Meta.”
Safeguards Built Around Autonomous Actions
With AI agents receiving greater access to business systems and information, control over what they can do has become a key consideration for companies.
OpenAI said Dots allow users to create custom rules governing what the agents can do and when they need permission. Actions involving sensitive operations, including changing passwords and permanently deleting data, require explicit user consent.
The company also said Dots operate on their own cloud computers and that private intelligence tools are being introduced to identify potential safety risks without retaining business customers’ data.
OpenAI reiterated that business customer data is not used to train its AI by default. Personal-plan users can choose whether their conversations and Dots work are shared for training.
Enterprise AI Race Intensifies
The launch comes as OpenAI faces increasing competition in autonomous AI, including from Meta. OpenAI is also continuing to expand the use of its business-focused products.
The company said weekly users of Codex and ChatGPT Work have surpassed 35 million, compared with about 5 million Codex users in June. Consumer-focused ChatGPT, meanwhile, has more than 1.2 billion weekly users, according to OpenAI.
For OpenAI, Dots represent another step toward artificial intelligence (AI) systems capable of independently handling complex, ongoing work. For businesses, the rollout also highlights the growing need to balance autonomous AI capabilities with clear controls around data, permissions, and sensitive actions.