Artificial Intelligence
Written by: Tathagata Sen
Updated 6:49 PM EDT, September 29, 2026

Photo credit: Unsplash.com
OpenAI on September 29 introduced Dots, a new type of AI agent that lets users delegate ongoing work to AI agents that can continue working without constant prompting. The company announced Dots at its DevDay event in San Francisco.
According to a post by OpenAI, Dots are designed to handle recurring, multi-step tasks on their own, allowing users to delegate work rather than start a new interaction for every step. They can be accessed through ChatGPT, text messages, Slack or voice and can work across connected applications.
OpenAI is introducing Dots to help businesses move from on-demand AI assistance to continuous task execution. Users can assign Dots specific responsibilities, with the agents continuing to work and returning when they need a decision or additional input.
Unlike a conventional chatbot that waits for a new prompt, Dots can continue working on assigned tasks and return to users when they need a decision. OpenAI is rolling out the feature to eligible business and enterprise users, with enterprise workspace controls for managing access and capabilities.
Enterprise administrators can control who can use Dots and whether they can connect to Slack and Microsoft Teams. They can also control whether Dots can access local computers and whether users can create custom rules that guide their actions and confirmations.
For chief data officers (CDOs), the shift toward always-on agents raises questions about access, ownership and monitoring. A Dot that can work across connected applications may interact with business information over an extended period, making it important to define what data it can access and what actions it can take.
CDOs can also help establish AI governance around agent permissions and accountability. That includes defining who can deploy agents, reviewing the systems and data they can access, monitoring their activity and ensuring that access changes when an agent’s role changes.
OpenAI’s enterprise controls already point to these governance needs. Dots access is off by default for Enterprise, and administrators can manage access through workspace permissions and custom roles.