Meta готовит рывок в персональных AI-агентах: план Цукерберга с отчёта за Q2 2026

Meta on July 29, 2026, at its earnings call for the second quarter of 2026, had CEO Mark Zuckerberg announce plans for a major push into personal AI agents — services that will be able to act around the clock on behalf of the user and help with goals, health, relationships, and finances.

What Zuckerberg promised

Mark Zuckerberg, CEO of Meta, described future agents as programs that "work 24/7 for you" and take on real actions rather than simply answering questions. According to him at the conference on July 29, 2026, such an agent should help the user achieve goals and improve their life — from health and relationships to finances. Meta did not name specific products, launch dates, or prices: Zuckerberg outlined only a high-level vision as part of the "all-in on AI" strategy. The statement was made at the quarterly earnings call for investors — a platform the company has used to position AI as its top priority.

Key facts from the statement:

• Made on July 29, 2026, at the earnings call for the second quarter of 2026
• Agents are designed to work 24/7 on behalf of the user
• Named domains of application: goals, health, relationships, finances
• The first area where agents "really took off" is programming
• Meta disclosed neither products, nor timelines, nor pricing

"Soon we'll have agents that can work for you 24/7, helping you achieve goals and improve your life, health, relationships, finances — whatever it is.

The first area where agents have really taken off is coding."

— Mark Zuckerberg, CEO of Meta, at the Q2 2026 earnings call

What is a personal AI agent?

A personal AI agent is a program that performs tasks on behalf of the user, rather than simply giving a text answer like an ordinary chatbot. That is exactly how Meta framed the concept at the conference on July 29, 2026: the agent "does something for you." The difference is fundamental — instead of advice on how to book a doctor's appointment, the agent books it itself; instead of reviewing a budget, it manages the payments itself. As Zuckerberg envisions it, this should work continuously, 24 hours a day, without constant commands from a human.

Zuckerberg named programming as the first proven case of such autonomy: it is precisely in writing code that agents have already become a mass-adopted tool. This confirms that the technology works in a specific niche — but it raises the question of transferring it to everything else, where the user is less prepared and the cost of error is higher.

Why mass rollout is a challenge

The main barrier, as Zuckerberg himself named it, is convincing less technical people to try agents at all. So far, according to him at the conference on July 29, 2026, autonomous agents have taken hold mainly in programming — an environment where users are technically savvy and willing to tolerate mistakes. Transferring the same model to a mass audience that will entrust an agent with health and money is harder: it requires a different level of reliability and trust. Meta has not yet explained exactly how it will solve this problem or on which of its platforms it will roll out agents, although the company has obvious distribution channels — billions of users of Facebook, Instagram, and WhatsApp.

What this means

Personal agents are becoming the next battleground for competition among major technology companies, and Meta is publicly staking its claim to this direction. But the statement made on July 29, 2026, is a vision, not a product: without specific timelines, prices, and working services, it is too early to judge a real breakthrough. The nearest benchmark is whether Meta can show not a declaration but the first agent that a mass-market user will trust with at least one real task.

• Meta is recognized as an extremist organization and is banned in Russia.