Know Your Agent: Building trust in an agent-driven world

That’s right. There’s so much buzz around agentic commerce right now. Not just chatter, you can see that people are enthusiastic about using this new technology to find the right prices, the right items, and generally make their online shopping experience easier and more streamlined.

In fact, Experian’s annual Identity & Fraud report found that nearly a third (31%) of consumers report that they have already used AI to shop or transact online, and another 23% would consider doing so.

That’s one surprise. It’s taking off quickly. It feels like a natural next step to using Google to search, and to the personalization trend of the last several years. Why not use GenAI?

But the catch is, currently, it mostly stops at research such as identifying the best item, price comparison, that kind of thing. Not many people are willing to use AI all the way through to start handling purchases in an automated way.

Even those who will, are generally only willing to consider small-dollar transactions. Our report found consumer confidence drops sharply for higher-stakes decisions with just 21% comfortable relying on AI for completing travel-related purchases and only 17% for financial services decisions.

This was a core part of our research. What we found is that it comes down to trust.

Right now, there’s a trust issue with handing over transactions to agents.

What’s really interesting is that this lack of trust isn’t down to mistrust of the technology, which might have been something you’d expect. It’s a lack of a trust framework.

When we make transactions online today, we understand the framework we’re operating in, no matter whether we’re looking at this from the perspective of a customer, a merchant, a card network, or a bank. We know the rules, we know what happens if something goes wrong – and it does sometimes, and we know that too.

Once an agent is in the mix, no one knows what the rules should look like yet. Who is making the transaction? How do you know if the agent is authorized? What happens if the wrong item is delivered, or if an agent is compromised? It’s this structural hole that is holding up participation.

Exactly. Which is precisely the link we’re building with Experian Agent Trust™ and all the partnerships across the ecosystem that are part of it.

Agents represent a new player in the online landscape, and we need to be able to understand them, verify them, and incorporate them into the systems that we use and rely on every day.

What I think people don’t yet realize is that this goes well beyond agentic commerce.

It already is. Any CISO today will talk to you about the IAM challenges inherent in agents. Who is taking a given action? Who does this agent belong to, and what was it created to do? What access does this agent have? Is it auditable?

All these questions are more pertinent if it’s your job to protect an organization’s cyber posture. All these questions come down to agentic trust.

Then there’s HR, both in terms of recruitment and in terms of employee culture and safety. There’s privacy, for employees and organizations but also for consumers. There’s the gamut of activities where you might engage an agent, which is already broad. I think OpenClaw was a very cautionary tale for a lot of people in that respect. Trust is crucial and non-negotiable. Without it, AI will not scale.

I am seeing enthusiasm, even more than perhaps I would have anticipated six months ago.

People want to be more efficient personally and professionally, to be able to spend more time on the strategy of their work, and spend less time on life admin. They want to keep up, as well, and not get left behind technologically speaking.

At the same time, the trust message comes up again and again. People want transparency, so that they can trust it.

If there are going to be agents deputized to do things for you, as a professional or as a private individual, you want to know how many, and where they are, and what they’re doing on your behalf.

That’s a good place to start thinking about it, but then you have to take it much wider. This goes well beyond agentic commerce, remember.

If we’re looking at a world in which agents can set up bank accounts, or submit applications, or roam the online world looking for deals we said we wanted to take advantage of if they appeared any time in the next six months, then we’re going to want to know what our agents have been doing. Where have they been?

Consumers are going to want to be able to see that easily and reliably and transparently. Maybe even have alerts set up, in case an agent does something they weren’t expecting. That kind of transparency is the basis of agentic trust.

You can’t Know Your Agent if you don’t know where it’s been or what it’s been doing. That’s why Experian is creating this framework, as part of the integral foundation that will make agentic trust possible.

We’re already working with them, as you can see from the way Experian Agent Trust™ was launched in partnership with Visa, Cloudflare, Akamai, and Skyfire. But I envision this going beyond that. It’s really an area where you need to have the whole ecosystem joined together. This requires different groups and pools of collaboration, overlapping.

There’s the banks and payment rails, so that things run smoothly and identification can be pulled in whenever necessary. Then there’s regulators, shared standards, cybersecurity, and commerce.

Validation will require the input of experts in different fields, like the way we work with Incode to verify and connect identity elements such as government-issued IDs, facial recognition, liveness checks, and real-time metadata.

Beyond that, there’s our professional lives and the services and systems we use in them, and the events and interactions that make up our personal and social lives as well.

Really, this is about trying to build a framework that removes the confusion, and supports the emergence of agents into any aspect of our world, so that the support is in place as the use cases emerge.

Right now, we’re all trying to run and build at the same time things are being created. It’s no wonder that there’s hesitation and uncertainty at such a dynamic moment.

My perspective is that this technology has huge potential to uplift and empower so many elements of our lives and our world. If we put the trust framework in place, we’ll be able to see that potential turn into actuality, safely and securely.

AI brings enormous capability, and it demands immense security to match.