OpenAI and Meta Bring AI Agents Into Connected Apps

OpenAI and Meta are bringing AI agents into connected apps, where they can carry out tasks after receiving a goal. Both companies describe controls for app access and approval of some actions.

OpenAI introduced dots on Sept. 29. The agents run on GPT-6 Astra, use their own cloud computers and can connect to more than 4,000 apps through OpenAI’s plugin ecosystem. Users can access them through ChatGPT, Slack and Microsoft Teams. Rollout is beginning for eligible Pro, Business Premium and Enterprise plans.

Meta launched Muse as a personal AI agent on Sept. 8 and expanded it for small businesses on Sept. 29. The agent runs on a dedicated virtual machine and can work across connected services, including business tools and Meta accounts. In the small-business offering, approval is required before Muse publishes, sends or spends.

Users can select which apps OpenAI’s dots access and set rules to allow, block or require approval for specific actions. Password changes remain under the user’s control. Muse users also choose connected apps and their permissions; the agent may ask for approval before sending email or making a purchase.

The companies describe separate controls for data used to improve AI models. Content from ChatGPT Business, Enterprise and Edu workspaces is not used to improve OpenAI’s models by default. Users on personal plans can control whether their dots conversations and work are used for model improvement. Users can opt out of using Muse interactions to train Meta’s AI models; those interactions are not shared with Meta’s advertising systems.

Meta says it plans to introduce a Confidential VM later in 2026. The product announcements provide no comparative performance data or independent assessment of how the safeguards work in everyday use.