OpenAI Launches 'dots', An Always-On Personal AI Assistant Evolving from Coding Tools

Image: TechCrunch · Source

OpenAI's new personal assistant 'dots', available to ChatGPT Pro users, proactively helps manage tasks by integrating with users' calendars, emails, and more. It builds on earlier coding tools but aims for a seamless, intuitive experience.

OpenAI has introduced "dots," an always-on personal AI assistant designed to work continuously in the background, enhancing user productivity by anticipating needs without constant prompting. The assistant debuted at OpenAI’s developer day in late September and is now accessible within ChatGPT desktop and mobile applications for subscribers of the Pro tier and above, which starts at $100 per month.

Alexander Embiricos, who leads OpenAI’s enterprise product team, explained that the genesis of dots was OpenAI's desire to transform complex AI interactions into a single coherent conversation with a persistent entity rather than fragmented sessions. This approach aims to make interactions more intuitive and human-like.

Building on OpenAI’s earlier Codex-powered coding tools, dots represents a repackaging of existing model functionalities into a broader personal assistant that integrates with users’ calendars, email, and computers to assist with everyday tasks. Embiricos noted that while Codex allowed some automation, dots reduces user effort by staying active continuously, even when no direct input is given.

The product’s name, dots, plays off the negative space in OpenAI’s logo, chosen to convey ubiquity and simplicity. This persistent agent model emphasizes proactive assistance, such as alerting users to calendar conflicts or deadlines. Embiricos acknowledged some early timing issues with notifications but expects improvements over time. He also shared an example of dots detecting and analyzing a system outage before engineers became aware, demonstrating its potential for real-time problem-solving.

Privacy remains a priority, with OpenAI incorporating "proactive research," where dots periodically evaluates available information to enhance its help while respecting user permissions. Users maintain control over data access and can disable model training updates to prevent data from being used to improve OpenAI’s AI models. However, the controls are currently broad, and the assistant requires significant access levels, such as refusing read-only cloud drive access.

Embiricos recognizes the challenge in balancing helpfulness and user control, noting conservative default permissions at launch and commitments to develop more granular privacy settings going forward. He also highlighted industry concerns about scalability and user acceptance, mentioning that despite AI agents’ capabilities, people and businesses often prefer human interaction for complex engagements.

Future developments may include deeper collaborations with payment processors to enable assistants like dots to perform transactions on users’ behalf, though this remains sensitive due to privacy and security considerations. For now, OpenAI focuses on creating a smart, discreet, and useful assistant that seamlessly supports daily workflows without overwhelming users or backend infrastructures.

OpenAI’s dots exemplifies the evolving landscape of AI assistants, shifting from command-based tools to persistent entities aiming to anticipate and proactively support user needs, marking a significant step in AI integration into everyday professional life.