Mark Zuckerberg, CEO of Meta, noted in a recent statement that progress in developing AI agents within the company has fallen behind planned timelines. This admission comes amid growing market interest in autonomous digital assistants, but it appears that even industry giants are encountering unexpected technical and operational challenges.

According to Zuckerberg, over the past four months, the AI agent direction has not accelerated as Meta had expected. The company is investing significant resources in this technology, yet results have not matched ambitious internal forecasts. This statement underscores that even with powerful infrastructure and multi-billion dollar budgets, creating truly effective and scalable AI agents remains a non-trivial task.

Interestingly, despite the slowdown at Meta, the ecosystem around AI agents continues to develop. According to data from the analytics platform Artemis, the volume of transactions through the x402 protocol, which supports the operation of such agents, amounted to about $2 million over the past 30 days. This indicates that while major players are adjusting their expectations, niche protocols and solutions are finding their audience and demonstrating practical value.

My analysis: Zuckerberg's admission is not so much a signal of problems as a realistic assessment of the current state of the technology. The AI agent market is overheated with expectations, and actual implementation always lags behind the hype. Meta, like many others, is going through the "valley of disappointment" characteristic of any breakthrough technology. However, $2 million in turnover through x402 in a month is a modest but real indicator that the first practical use cases already exist. Investors should pay attention not to loud statements, but to specific usage metrics.