Your Contact Centre May Soon Answer to a Customer's AI, Not the Customer
In 2026, Meta CEO Mark Zuckerberg published an essay on what he calls "personal superintelligence," announced on X, arguing that individuals should get their own AI agents to pursue goals, manage information and eventually act on their behalf. CX Today, which broke coverage of the essay, reads this as a proposal that a customer's personal agent could filter brand communications, assess offers and handle routine service requests before a human ever sees them. The outlet also spoke to Jeff Fettes, CEO of contact-centre AI vendor Laivly, about what that shift would mean for support leaders already deploying their own bots.
What is Zuckerberg actually proposing?
Meta's founder wants individual users equipped with personal AI agents that filter incoming brand messages, evaluate offers and manage routine interactions automatically on the customer's behalf.
The essay frames this as empowerment: rather than superintelligence sitting inside a handful of institutions, Zuckerberg wants it distributed to individuals for tasks spanning education, health, creativity and everyday admin. Customer service is a small slice of that ambition, not the focus of the essay, but it's the slice that should worry anyone running a support desk — because it implies a new layer of automated gatekeeping sitting between your brand and the person you're trying to reach.
Why should support software buyers care right now?
If AI agents start pre-screening brand outreach and service requests, contact centres will need machine-readable policies, reliable APIs and clear permissioning long before adoption is widespread.
That's a familiar problem for anyone who's already had to rethink self-service around low-code tooling or questioned whether IVR menus still serve anyone, as this site has covered before. The difference here is the direction of travel: instead of brands building the AI layer their customers interact with, an outside platform — Meta, in this telling — builds it for the customer, and brands have to make their pricing, availability and policy data legible to a system they don't control.
How much of this should you trust as a roadmap rather than a pitch?
Treat the essay as a positioning statement, not a shipped product — Zuckerberg defines "superintelligence" and "empowerment" himself, with no independent measure offered.
Zuckerberg says he intends everyone to have access, telling followers "I believe everyone should have access to superintelligence." That's a values statement, not a spec, and it arrives from the same company managing its own headcount debates — CX Today notes Meta's own recent backtracking on layoffs undercuts any tidy story about AI simply replacing labour. Fettes made a related point to the outlet: efficiency should be a byproduct of a good AI strategy, not the reason a company adopts one.
What should you ask your own AI vendor as a result?
Ask exactly how a personal agent would authenticate, what data it can pull through your APIs, and who is liable when it acts wrongly.
None of that exists as a standard yet, which is precisely the point — if this direction takes hold, the vendors who win will be the ones who can show a working handshake with customer-side agents, not just another internal chatbot demo.
Frequently asked questions
Has Meta released a product based on this essay?
No. It's a philosophy essay Zuckerberg posted to X in 2026 setting out intent, not a feature launch or a tested integration.
Does this replace the case for IVR or chatbot investment?
Not on current evidence — it's a proposed future layer on the customer's side, separate from the self-service tooling brands already run.
Who is Jeff Fettes and why is he quoted on this?
Fettes is CEO of Laivly, a contact-centre AI vendor; CX Today asked him to react to the CX implications of Zuckerberg's essay.
Source: CX Today.