Meta’s next big move: AI assistants for everyday life — what we know so far

In Meta’s sprawling AI labs, the next big bet is not another app feature or social feed tweak, but something far more intimate: an assistant that could know its users well enough to act on their behalf.

The company is reportedly building a personalised AI agent designed to carry out everyday tasks for its billions of users, marking one of its most ambitious attempts yet to embed artificial intelligence across its ecosystem. At the centre of this effort is a new model, Muse Spark, which will power an advanced digital assistant currently being tested internally by employees.

According to people familiar with the project, Meta’s goal is to create an AI system that behaves less like a chatbot and more like a proactive digital companion. One insider described it as similar in spirit to systems like OpenClaw, which allows users to build autonomous agents capable of completing tasks without constant supervision.

More from Tech

Meta’s AI assistant push, powered by Muse Spark

The assistant under development is expected to sit at the centre of Meta’s broader agentic AI strategy, designed for its more than 3 billion users across platforms. Unlike traditional assistants that respond to prompts, Meta’s vision is to build tools that can anticipate needs and execute actions, from scheduling and communication to more complex personal tasks.

A defining feature of the project is its ambition to allow users to share highly sensitive data, including health and financial information, in order to enable deeper personalisation.

While this could significantly expand the assistant’s usefulness, it also raises immediate questions around privacy and trust. One person close to the project cautioned that “there’s a trust deficit as wide as the Grand Canyon”, reflecting uncertainty over whether users will feel comfortable granting such access.

Internally, the assistant is already being tested by a select group of staff as Meta refines its capabilities ahead of a wider rollout.

Zuckerberg’s “personal superintelligence” vision meets investor pressure

The project is a direct extension of Mark Zuckerberg’s long-term vision of integrating AI deeply into everyday digital life. He has repeatedly spoken about building “personal superintelligence” systems that can act as extensions of individual users, reshaping how people interact with technology.

However, the scale of that ambition is increasingly drawing scrutiny from investors. Meta has significantly ramped up spending on AI infrastructure and talent, even as it prepares to cut around 10 per cent of its workforce. The company recently raised its capital expenditure forecast by $10 billion, bringing it to as much as $145 billion for the year.

Markets reacted sharply, with Meta’s valuation reportedly shedding around $170 billion following concerns over escalating costs and unclear timelines for returns on AI investment.

Alongside the assistant project, Meta is also developing AI-driven 3D characters capable of real-time interaction, including a digital version of Zuckerberg himself, underscoring just how central AI has become to the company’s product roadmap.

Yet the tension remains clear. On one side is Zuckerberg’s conviction that AI will redefine Meta’s future. On the other is a growing demand from investors for clearer evidence that this expensive transformation can deliver measurable returns.

For now, Meta is betting that users will eventually trust an AI that knows them well enough to act for them. Whether that trust arrives as quickly as the technology is being built remains the unanswered question.