Zuckerberg tells Meta staff AI agents are progressing slower than planned after restructuring shake-up

Meta Platforms CEO Mark Zuckerberg told employees at an internal town hall that the company’s work on AI agents has not progressed as quickly as he had planned. The remarks, reported by Proactive Financial and other outlets, point to frustration inside the company with the speed of technical delivery even as Meta attempts to reorganize around artificial intelligence.

The report characterizes Zuckerberg’s message as a direct acknowledgement that Meta’s sweeping restructuring efforts have not yet translated into the expected acceleration for agentic systems. Those AI agents are software designed to carry out tasks with some level of autonomy, rather than simply generating text, and Meta has positioned agent development as a central part of its AI strategy.

Proactive, citing the internal meeting, said Zuckerberg told staff that the progress of Meta’s AI agent technology has been slower than expected. Several other technology and finance outlets also reported similar framing that the agent-development timeline is not meeting earlier expectations, though specific performance metrics, target dates, or system benchmarks were not included in the available reporting.

The comments land as Meta continues to reshape its organization for the AI era. Separately, CNBC reported in April that Meta would cut about 10% of its workforce as it pushes deeper into AI. That kind of cost and staffing change can be used to reallocate resources toward new product priorities, but it can also create execution risk if key engineering milestones slip.

Meta has not, in the reporting available here, provided additional details on what exactly is behind the slower pace. The internal nature of the town hall also means employees were not necessarily given public, verifiable updates on what has changed in the underlying systems, what parts of the roadmap are delayed, or whether the company is revising schedules for customer-facing launches.

The situation also underscores a broader sector challenge. Across Big Tech, “agentic” AI has been marketed as a major step beyond chat, but building reliable systems that can take actions safely and usefully in real workflows is technically difficult. When companies reorganize to prioritize agents, they typically face bottlenecks spanning model capability, tooling integrations, and safety layers, any of which can affect delivery speed.

For investors and product teams, the most consequential unknown is timing. The reporting does not specify whether Meta is behind on internal demos, external deployments, or either. It also does not indicate whether the slower pace reflects engineering trade-offs, higher quality or safety constraints, or a shift in priorities following the restructuring.

What to watch next is whether Meta addresses the pace publicly, for example through subsequent product announcements, developer documentation updates, or forward guidance around agent-related releases. Without additional disclosures, the town hall remarks primarily announcement that the company’s internal execution on agentic AI is still under pressure, even as leadership sets an aggressive AI agenda.