OpenAI has outlined how its always-on AI assistant, dots, can independently monitor tasks, research information and provide recommendations without waiting for user instructions. Speaking at TechCrunch Disrupt 2026, OpenAI product leader Alexander Embiricos discussed the assistant’s proactive capabilities, privacy safeguards and challenges associated with giving AI systems greater autonomy.
OpenAI introduced dots on September 29, 2026, as a personal AI agent designed to work continuously across connected applications and services. According to the company’s official announcement, the assistant uses its own cloud computer and can connect to thousands of applications through plugins, enabling it to support ongoing tasks.
Dots Uses Personal Context to Anticipate Tasks
Unlike traditional chatbots that respond to individual requests, dots can examine permitted information from calendars, emails and connected applications to identify tasks requiring attention. The assistant uses a feature called proactive research to periodically review available context and suggest actions that could help users.
Embiricos said the approach builds on capabilities previously developed for OpenAI’s Codex coding assistant and ChatGPT Work. Instead of requiring users to manually initiate individual tasks, dots presents those capabilities through a persistent AI assistant that can learn from feedback.
Early testing has also revealed limitations in the assistant’s ability to provide timely notifications. Embiricos recalled receiving an alert about an upcoming appointment only three minutes beforehand, illustrating the need for better scheduling and prioritization.
He also described an example involving an engineer at a drone delivery startup whose AI assistant reportedly detected a system outage, analyzed the problem and presented possible solutions before the employee noticed the issue.
Privacy Controls Remain a Central Challenge
The assistant’s reliance on personal information raises questions about how much access users should provide to autonomous software. Dots can use connected accounts and conversation history to personalize recommendations, although its functionality depends on the permissions granted.
According to OpenAI’s privacy documentation, users can manage connected services, establish rules governing permitted actions and disable the use of eligible personal conversations for model training. Proactive research can access authorized information but cannot independently send messages, modify connected files or control computers.
Embiricos acknowledged that more detailed permission controls are necessary, particularly when AI assistants interact with sensitive accounts. He said OpenAI initially adopted conservative defaults requiring frequent user approval, with adjustments planned as the technology develops.
OpenAI Examines Payments and Wider AI Agent Adoption
Embiricos also discussed the challenges of integrating autonomous assistants with businesses, websites and financial services. He said OpenAI is holding discussions with payment processors about potential AI-assisted transactions but did not disclose specific agreements or launch plans.
The company also faces questions about whether organizations will accept interactions initiated by automated assistants. Embiricos said he continues to rely on human assistance for professional communications because business partners often prefer dealing directly with people.
Dots initially launched for eligible Pro, Business Premium and Enterprise subscribers, with OpenAI restricting availability while evaluating safety and performance.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.