Optimists, Agents, and Actual Fireworks: Notes from Alloy Summit 2026

Earlier this month, over 120 senior fraud, risk, business, and product leaders from more than 75 banks, credit unions, and fintechs spent a day and a half together in Chicago for the Alloy Summit. This year’s theme was Meet the Moment: Navigating Risk in the AI Era.

On day one, we polled the room and asked how they were feeling about AI's impact on their role. 72% described themselves as “AI optimists.” Coming from a group whose entire profession is imagining what could go wrong, I'll admit that surprised me.

Fraud is moving faster than rules can be rewritten, customers are beginning to hand their accounts over to AI agents, and nearly everyone is being asked to do more with less.

If I had to name the single idea that ran through everything — the keynote, the product session, the practitioner panels, the conversations over coffee — it's this: trust is the most valuable asset any of us have, and it now has to be earned continuously.

Chris Bashinelli, a National Geographic Explorer, opened the Summit by reminding a room full of technology leaders that trust still comes down to human connection. He made the point through stories and video from his travels around the world, and brought the room to laughter and a few tears in the process. His argument was that as AI absorbs more of the technical work, human connection doesn't become obsolete. It becomes scarce, and therefore more valuable.

Michelle Beyo of Finavator put the same idea in operational terms: "Trust is the new currency — earn it in real time." Identity isn't a single gate at the front door. It's a signal that runs across the entire customer relationship.

Here's what that looked like in practice across two days.

The hard part of AI isn't the AI. It's the process and the people.

The most consistent piece of advice from the stage had nothing to do with models.

Martha Cummings, a Marqeta board member, and our co-founder Laura Spiekerman landed on the line that got repeated all week: put AI on a bad process, and you get a bad process on steroids. The upside is that AI becomes a forcing function. You find out what's actually broken.

Other panelists confirmed this idea. 

A digital banking executive at a large regional bank advised those in the room to "slow down to go fast." Clean data and well-documented processes aren't nice-to-haves. They're prerequisites. A chief banking officer for a tech-forward, national small-business lender offered a useful reframe: Don't start with "Where can we use AI?" Start with "Where is a process taking a measurable toll?" Then reimagine that process end to end.

The people side of things turned out to be just as decisive. That optimism I mentioned earlier isn't the same thing as readiness, and several leaders made the point that the skill worth investing in isn't technical. It's judgment. Knowing which what-if questions to ask. Spotting when an output doesn't make sense. Anticipating how a customer will experience it. The technical parts get learned; the judgment parts require an actual human touch. 

Customers are starting to bring agents with them

When we asked how many people in the room were already using AI agents, at work or in their personal lives, about half the hands went up.

Which is why Know Your Agent (KYA) came up again and again over the two days, from Alloy speakers and industry speakers alike. 

Grace Liu, Senior Product Director at Alloy, walked through what that means for authentication with an example that perfectly captured the problem: an agent acting on a customer's behalf lands on a webpage carrying an instruction that overrides its own. There was no bad intent, no malicious behavior. The credentials checked out the entire time.

Static authentication can't catch that. It can confirm who someone is, but it can't notice that the behavior stopped making sense. Which is exactly why the question has to shift from "Who are you?" asked once in the beginning, to "Does this still make sense?" asked at every step of the journey.

Customers want agents to make their lives easier — to pay the wedding vendor on time, to pay down a credit card balance twice a month, to move money into savings. If you can't find a way to let them do that safely, they'll find someone who can.

Nobody can see the whole picture alone

Alloy co-founder and CEO Tommy Nicholas made the argument that reframed the arms race for a lot of the room. Everyone has access to the same AI now, fraudsters included. The difference is that nothing holds their systems back, and ours have guardrails. Which means writing better rules isn't the answer anymore. No matter how good they are, static rules can't keep up.

So if better rules aren't the edge, what is? Tommy's answer was being able to see more than any one institution can see on its own. That's what Fraud Signal, Alloy's predictive machine learning model, is built to do. Fraud signal looks for fraud patterns across the Alloy network to help financial institutions and fintechs identify trends sooner. A scam that's brand new to your institution may already be well understood elsewhere on the network. His push to the room was to put that intelligence to work across every workflow, not just onboarding.

Our panel on rethinking detection, with Plaid, Inscribe, and Nasdaq Verafin, made a related point: stop trying to find the needle and start shrinking the haystack. Cut false positives first, then make the investigation itself faster.

We announced our new partnership with Nasdaq Verafin from the stage, unlocking access to even more shared intelligence across the customer lifecycle for our mutual clients. Fraud insights from Nasdaq Verafin’s consortium data network of over 850 counterparties now flow into Alloy, while Alloy’s real-time fraud alerts feed into Nasdaq Verafin’s investigation and case management platform. Case outcomes from Nasdaq Verafin then flow back into Alloy’s Fraud Signal model, improving end-to-end fraud controls with every resolved case.  

The best parts of an event happen off-stage 

We closed the Alloy Summit by asking attendees to name their favorite part of the last two days. No surprises here – it was the people, by 58%. 

This is a great reminder of why senior leaders choose to leave their organizations, their families, and their day-to-day responsibilities to spend time together in person. You can read a panel recap from your desk. What you can't do is compare notes with your peers, or hear how a similar institution solved the problem currently keeping you up at night. Our cruise on the Chicago River and the surprise fireworks show made a strong case for showing up, too.

Thank you to everyone who traveled to Chicago, to the speakers and partners who shaped what happened on stage, and to the Alloy team who built this event.