Siri's founder says 'We have the opportunity to design it right this time, so that AI is on our side'

Tom Gruber told LIVEX 2026 in Abu Dhabi how personal agents could reshape city services

Last updated:
Areeba Hashmi, Reporter
Siri's founder says 'We have the opportunity to design it right this time, so that AI is on our side'

Dubai: Fifteen years after Siri launched, the man who founded it thinks the assistant's original promise is only now being fulfilled, and that cities are where it gets interesting.

Tom Gruber, founder of Siri and Humanistic AI, was the first speaker at "The city that thinks" at LIVEX 2026 at ADNEC Centre Abu Dhabi, moderated by Dr Ayesha Khanna, CEO of Addo AI. His case was that a personal agent is not just a faster way to fill in forms. It is a way for residents to claim what they are owed without handing over their private lives.

The vision, completed

Asked about the wave of personal AI agent launches, Gruber drew a line between agents built for work and agents built for a life.

"Personal AI is different than commercial AI and workflow," he said. "Personal AI knows everything about your life and needs to do a really good job of being your personal system."

Of the latest arrivals, he was generous. "It does effectively complete the vision of Siri."

That vision, as he described it, was simple enough: tell the system what you want in plain language and it gets it done. What has changed is reach. An agent can now do the things a person does with their fingers on the web, including forms and phone calls.

Why a city is the harder problem

Gruber's interest is in what comes after that.

"There's a lot more that AI agents can do than all the main things that we already do," he said. "We already fill out forms, we already call people."

A city runs an enormous number of services, each with its own rules and protocols, and no resident can hold all of it in their head. An agent can. That, Gruber argued, changes what is discoverable in the first place, because the question stops being what you happen to know about and becomes what actually applies to you.

Proving you qualify, without showing why

His example was a family member with a chronic condition who needs frequent hospital visits. A city might grant people in that position priority in its transit system, or reserved parking.

Today, claiming that means one thing. "You would show your medical record to the city government."

In the version Gruber described, the agent already helps manage the relationship with the doctor. It can see a certificate signed by that doctor confirming the condition, and it can read the city's own published rule setting out the entitlement. It then proves eligibility cryptographically. The city attaches the benefit to the resident's digital transit card and never learns the diagnosis.

The implication, he said, runs well beyond health. Means-tested services that depend on income work the same way, and many city services currently demand personal information as the price of access.

"Future AI agents will be able to be our advocate, give us what we're entitled to without revealing personal information."

Getting the design right this time

Khanna asked him to explain humanistic AI, and why it matters more for cities than for companies.

Gruber's answer began with a warning drawn from the last technology cycle. "The internet was designed with some flaws," he said, and those flaws led to the social media takeover and its effects on society.

Cities, he argued, are not stuck with that inheritance. "We have the opportunity to design it right this time, so that AI is on our side."

Humanistic AI, in his definition, is AI built by design and on purpose to be on the side of the human. The measure of success is the joint performance of the human and the machine working as a team, rather than a machine competing with a person on the same terms.

He sketched what that looks like for a city. A city delivers transport, education, health and energy, and it wants to improve at all four. Its users are its residents. Put the two together, he said, and you have a learning system, in which verified users of a service can have their agents pass feedback back to the city automatically, with a guarantee attached.

"It's not a bot faking it out. It's a real human."

The contract underneath all of it, in his framing, is what people are really deciding now. "It doesn't have to be big platform in California owns everything," he said. The alternative is that everyone has powerful AI working for them personally.

Get Updates on Topics You Choose

By signing up, you agree to our Privacy Policy and Terms of Use.
Up Next