Hands-on with Dots, OpenAI's work-focused agentic product: highly capable and intuitive, with natural-feeling conversations represented as a chat inside ChatGPT

Your AI assistant can now draft legal documents and research office building specifications, saving you hours of work.

OpenAI has launched Dots, a new AI agent designed for work-related tasks, integrated within ChatGPT. Unlike some competitors that are free or consumer-focused, Dots is available exclusively to paid ChatGPT users. The agent aims to streamline workflows by handling tasks such as drafting emails, researching information, and organizing schedules, drawing from user-provided data and documents. During initial testing, Dots demonstrated significant capability in assisting with complex work, including legal and financial document preparation, and content generation. OpenAI emphasizes that Dots uses an improved AI model, GPT-6 Astra, to enhance safety and reliability following the discontinuation of a previous version due to user deception issues. The company acknowledges the significant trust required for users to grant AI agents access to sensitive information and is working to reassure users about data security and agent behavior.

Dots represents a significant step towards the agentic AI future envisioned by tech leaders, moving beyond basic chatbot functionalities to perform complex, multi-step work tasks. Its integration within ChatGPT offers a familiar interface, contrasting with other emerging agents that use distinct messaging platforms. OpenAI's decision to position Dots as a premium, paid product suggests a strategy to monetize advanced AI capabilities, potentially creating a market segment for users willing to pay for focused, reliable work assistance, differentiating itself from the ad-supported or free models of competitors like Meta's Muse.

The launch of Dots, alongside other recent agent introductions like Meta's Muse and Instinct, signals an acceleration in the development and adoption of AI agents for both personal and professional use. However, challenges remain, particularly in consumer branding and establishing trust due to privacy and security concerns, as highlighted by OpenAI's recent AI model issues. The market is still grappling with how to best position and monetize these agents, balancing a wide range of use cases from wedding planning to hardcore software engineering.