Meta is testing a new personal artificial intelligence agent with employees that can carry out online tasks on their behalf, as the technology company expands its efforts to develop AI systems capable of taking actions beyond answering questions.
Codenamed Project Hatch, the AI agent is being made available internally to Meta employees for testing. It is designed to perform tasks including booking restaurants, ordering food, shopping online, managing calendars and emails, conducting research and finding services such as dog sitters.
Unlike conventional AI chatbots that primarily respond to prompts, Hatch is being developed to operate more independently. The agent can use its own virtual computer to navigate websites, complete forms and carry out digital tasks delegated by the user.
The system is also designed to be personalised. Users can customise elements such as the agent's name, personality and areas of focus. It is expected to build an understanding of individual preferences over time, while permissions can be used to control access to information and sensitive actions.
The internal testing aligns with Meta CEO Mark Zuckerberg's broader push towards what the company describes as personal superintelligence. Meta has outlined a vision in which individuals could eventually use AI agents that understand their goals and continuously assist them across areas including relationships, careers, finances, health, household management and hobbies.
Zuckerberg has previously said he expects personal AI agents to become widely used over the coming years. During Meta's second-quarter earnings call, he said the company sees consumer personal agents as a significant opportunity and expects messaging platforms such as WhatsApp to become important surfaces for interacting with multiple AI agents.
Privacy and security are also expected to form part of Meta's approach. The company has said personal agents would need safeguards that allow users to trust them with sensitive information and tasks. Meta has also outlined plans for private AI experiences where personal information can remain inaccessible to the company itself.
Project Hatch is currently being tested beyond Meta Superintelligence Labs, allowing a broader group of employees to experiment with the technology and provide feedback.
The development places Meta alongside other major technology companies working on agentic AI systems capable of carrying out multi-step tasks for users. The shift represents a broader industry move from conversational AI towards systems that can use digital tools and act on a user's instructions.
Meta has not announced when Hatch could become available publicly or disclosed official pricing for a potential consumer version.
Disclaimer: This article may include information derived from interviews, press releases, public statements, research, company communications and other publicly available or third-party sources. Such material may be summarised, paraphrased or contextualised for journalistic and editorial purposes. All rights in third-party content remain with their respective owners.