Openai Launches Dots, Always-on Agents For Personal And Work Tasks

Powered by GPT-6 Astra, Dots can work across connected apps, seek approval for sensitive actions and, eventually, operate in teams.

OpenAI announced Dots at its DevDay event on Tuesday, introducing always-on AI agents designed to pursue tasks in the background rather than wait for a new prompt at every step. Powered by GPT-6 Astra, a Dot can use connected apps and a cloud computer to work toward a user’s goals, according to TechCrunch and The Verge. Users can initially create one personalized Dot; OpenAI says it plans to let them use multiple agents in the future.

Table of Contents
  1. How Dots work
  2. Availability and safeguards
  3. What it could mean for tech teams
  4. Sources

How Dots work

Dots can use a browser and more than 4,000 supported apps through their cloud computers, according to Engadget and The Verge. Users can check a Dot’s work by opening that computer. Engadget also reports that a Dot can be connected to a user’s PC, allowing it to work there alongside them.

Users can communicate with Dots in ChatGPT, Slack and Microsoft Teams, with context carrying across those services, the reports say. Engadget says ChatGPT users can also call a Dot on desktop, web or mobile. Text messaging is planned; Wired reports that Pro users can join a waitlist for iMessage or RCS messaging. While working, a Dot can send progress updates, questions or requests for a decision.

OpenAI’s examples span personal and professional tasks. Wired describes a dinner-planning demonstration in which a Dot noticed a calendar conflict and proposed food-delivery options. Engadget reports an example in which a Dot prepared a forgotten invoice and sent it after the user approved. For developers, OpenAI suggests using a Dot to track customer feedback and work on fixes or features, according to TechCrunch and The Verge.

Availability and safeguards

The outlets give different accounts of launch eligibility. TechCrunch says Dots are starting to roll out to Pro and Business Premium users in eligible markets, while Wired says the initial rollout is for ChatGPT Pro subscribers. The Verge includes Enterprise users among those gaining access, and Engadget says Pro, Business, Premium and Enterprise users in select markets will get a primary Dot, while describing the rollout starting Tuesday as limited to Pro and Business Premium subscribers.

OpenAI says Dots have rules governing when they can act alone and when they must seek approval. Users can set additional boundaries, and The Verge reports an auto-review feature intended to check proposed actions against those rules. Engadget says tools used for Dots’ proactive research are restricted to read-only access, so that research cannot itself send messages or change app content. Wired cautions that connecting personal data to an agent still carries privacy and security risks, including unintended disclosure or manipulation by an online adversary.

What it could mean for tech teams

OpenAI is also pursuing workplace uses. Wired reports that CEO Sam Altman announced specialist Dots for tasks including accounting, email marketing and legal analysis. TechCrunch says OpenAI is working with Microsoft on integration with Agent 365 security controls. For teams considering these tools, the developer examples show the intended workflow: an agent could monitor feedback and prepare changes, while people set permissions and review actions that require approval.

Sources

This story was compiled by AI from the reports below. Read the originals for the full details.