The technology has recently moved into a broader testing phase among Meta employees, marking a shift toward software that can act independently on a user's behalf.
This development positions Meta to compete directly with rival "action-oriented" AI tools from Google, OpenAI, and Anthropic.
Unlike standard chatbots, Hatch is built to run in the background even when its app is closed, managing personal finances, health data, and schedules by connecting to external services like Spotify and Instagram.
The project represents a significant effort by Meta to turn its massive investment in AI infrastructure into a consumer product that generates direct revenue, potentially moving beyond its traditional advertising-based business model.
Users will be able to customize their agents' names, personalities, and specific areas of focus, with the system designed to learn from corrections.
To address security concerns, the memo states that Hatch will require user approval for sensitive actions, such as finalizing a transaction.
While Meta has not officially announced a release date or pricing, internal discussions and reports suggest the company may launch the agent platform within the coming weeks and has considered a monthly subscription fee for the service.