Meta is developing highly personalized artificial intelligence (AI) agents designed to handle users' daily tasks. The move is seen as a decisive bet on AI, which the company has positioned as its core growth pillar, at a time when concerns over profitability are mounting amid a recent surge in large-scale investment.
The Financial Times reported Wednesday, citing sources, that Meta is developing AI agents targeting more than 3 billion users. The agents are expected to be built on the company's own AI model, "Muse Spark."
According to the report, the project is currently being tested internally with employees. Some sources said Meta aims to deliver functionality similar to that of OpenClo, an open-source-based AI.
Meta is also accelerating another AI project involving 3D characters capable of interacting with users in real time. As a first step, a digital replica of Chief Executive Officer Mark Zuckerberg is reportedly being created.
These moves are viewed as part of a strategy to establish AI as Meta's core service. CEO Zuckerberg has been pushing a shift toward an AI-centered business structure under his vision of "personal superintelligence." While Meta plans to cut 10% of its total workforce this month, it is pouring massive funds into AI infrastructure and talent acquisition. Meta recently raised its capital expenditure (CAPEX) for this year by $10 billion to as much as $145 billion.
However, controversy has emerged over the use of personal information in building AI agents. Meta is considering linking sensitive information such as health and financial data with its AI assistant, but questions remain about whether consumers will accept this. "The trust gap is as wide as the Grand Canyon," one source said.
Meanwhile, Meta is also targeting the enterprise AI agent market beyond personal services. The company is concurrently developing advanced business-use agents to help entrepreneurs acquire new customers.
