By Ronda Cepeda, Director of Auto & Transportation, ValidiFI
A group of us from ValidiFI spent several days at Money20/20 in Las Vegas, splitting up to cover sessions, booth conversations, and meetings with lenders, payment providers, and partners. I’ve been to my share of industry conferences, but this one felt different. For years, the focus was on moving transactions online. Now the industry is automating the decisions behind them, and every one of us came home talking about that shift.
Not one person our team met was asking whether automation is coming anymore. Everyone wanted to know how to trust it. When a decision gets made in milliseconds and nobody’s looking over its shoulder, the data behind it has to be right. Here are the four things we keep coming back to, and how we’re helping our clients handle them.
Key takeaways from Money20/20 2026
Agentic AI dominated the conversation, but many large banks, marketplaces, and payment platforms are not yet letting outside AI agents transact.
Agent identity is the next frontier. Several companies are building registries that AI agents would need to join before they can act on a consumer’s behalf.
Real-time payments shrink the window to stop fraud, so risk decisions have to happen before money moves.
Automated decisions are only as good as the data behind them, and buyers want intelligence embedded in their existing workflows.
Agentic AI in payments: big excitement, real hesitation
Agentic AI was the topic I couldn’t get away from. These are AI systems that don’t just suggest a financial action. They actually take it: moving money, paying a bill, opening an account, or completing a purchase on someone’s behalf.
For all the excitement, there was just as much caution. Several of the largest banks, marketplaces, and payment platforms aren’t letting outside AI agents operate inside their environments or execute transactions yet. Honestly, I get it. When a consumer hands an agent their bank login or a one-time passcode, the line between authorized automation and account takeover, where a fraudster gains control of someone’s account, gets blurry fast.
What our team took away from those conversations: agentic finance will only grow as fast as trust does. Institutions need a reliable way to know the person is who they say they are, the account is legitimate and actually theirs, and the action was really authorized.
AI agent identity: proving who, or what, is authorized to transact
Some of my favorite conversations were about what comes next for identity. Several companies are building identity databases that AI agents would have to register with before they can transact, basically a common standard for which agents can be trusted. Others are working on ways to defend against deepfakes (AI-generated fake video, audio, or images) and to verify the agents themselves.
Nobody has landed on the right model yet, and there’s plenty of debate about how these registries should work. But you can see where it’s heading. Over the next decade, the industry will have to prove not just who a customer is, but who or what is acting for them, and what they’re allowed to do.
Here’s the catch, though.
An agent registry only answers part of that question. It won’t tell you whether the bank account behind a transaction is valid, open, owned by the right person, or likely to have a payment sent back unpaid. That piece still has to come from somewhere.
Real-time payments mean less time to catch fraud
Our team also spent a lot of time talking about real-time payments, which move money in seconds instead of days, and about payment systems that can connect and work with one another. Traditional payment networks are being linked with digital-asset rails, the networks that move cryptocurrencies and other digital assets. Cards funded by stablecoins, digital currencies pegged to a traditional currency like the U.S. dollar, are showing up at everyday merchants. And fraud checks and sanctions screening, which flag people and businesses on government restricted lists, are expanding to cover these newer payment types.
Faster is great for customers, but it leaves a lot less time to catch a mistake. Once funds settle instantly, you can’t wait for a returned payment to tell you something went wrong.
You have to know before the money moves.
Automated financial decisions need trusted bank account data
This is the part that ties it all together for me. Automated decisions depend on reliable, permissioned financial data, meaning data the consumer has agreed to share. An AI agent or a decision engine (software that automatically approves or declines applications and payments) can move fast, but if the account information it relies on is stale, incomplete, or wrong, it just makes bad decisions faster.
Across our team’s meetings, we kept hearing a preference for connected ecosystems over point solutions, single-purpose tools that work in isolation. Lenders, payment providers, and platforms want intelligence that plugs right into their existing origination (the application and approval process for new loans and accounts), funding, payments, and fraud workflows. Nobody wants one more standalone step that adds friction for the customer.
How ValidiFI helps close the trust gap
Every one of these challenges comes back to the same need: a trusted signal about the bank account at the moment a decision is made. That’s where ValidiFI comes in. We go beyond basic bank account verification to deliver account intelligence that helps our clients make better automated decisions.
New this fall: Compass. Connecting those signals is a big part of why we launched Compass. It brings account and payment risk signals together, so clients can see what a simple validation check might miss before they make a decision.
A trust signal for agentic and automated transactions
Whether a payment is started by a person, a workflow, or an AI agent, our account intelligence adds an independent check that the account is valid and the risk is acceptable before funds move.
Risk assessment before execution
In a real-time world, our insights help reduce ACH returns, failed payments, and fraud losses by flagging risk up front rather than after the payment settles.
Built for the full lending lifecycle
From application and approval through funding, repayment, and ongoing portfolio management, account intelligence supports greater funding confidence, fewer failed payments, and stronger automation rates. In lending, including automotive, that means fewer surprises after the deal is done.
Embedded where decisions happen
We integrate into the origination, payments, fraud, and lending platforms our clients already use, so intelligence arrives inside the workflow instead of as an extra step.
What our clients really want are results they can measure:
Frequently asked questions
What is an AI agent registry?
An AI agent registry is a shared database that AI agents would have to be registered in before they’re allowed to transact. It would typically record who operates the agent, who it acts for, and what it’s permitted to do, so banks, merchants, and payment platforms can check an agent’s identity and permissions before letting it move money. Several companies are developing registries now, and no single standard has emerged yet. A registry can confirm the agent, but not the bank account behind the payment. That’s where account intelligence from ValidiFI comes in, checking that the account is valid, owned by the right person, and likely to clear before money moves.
Can you verify a bank account without the consumer sharing their bank login?
Yes. ValidiFI offers both non-permissioned and permissioned options. Non-permissioned checks validate the account and assess its risk without asking the consumer for a bank login or passcode, which avoids the kind of credential sharing that makes platforms wary of AI agents. When a consumer chooses to connect their account, permissioned data adds more detail. Clients can use either one, or both, depending on the decision they’re making.
How does ValidiFI help reduce risk in real-time and agent-initiated payments?
ValidiFI checks that a bank account is valid, open, owned by the right person, and unlikely to return a payment before funds move. That account intelligence works the same whether a person, a workflow, or an AI agent started the transaction, and it plugs into the origination, payments, and fraud platforms lenders already use.
The bottom line
I came home from Money20/20 thinking trust is going to decide how fast all of this moves. AI agents, instant payments, and automated lending only work if everyone involved can rely on the data behind each decision. That’s the part our team cares most about, and it’s where we’re putting our energy.
If you’re figuring out how agentic AI, real-time payments, or automated lending will change your risk picture, I’d love to hear what you’re seeing.
Talk to our team
About the author
Ronda Cepeda is Director of Auto & Transportation at ValidiFI. She works with auto lenders, dealers, fleets, rideshare, rental, and other mobility companies to help them onboard customers faster, stop more fraud, and get paid without the headaches.
She has spent her career building partnerships with lenders, dealer groups, and fintechs, and she loves finding practical ways to make payments and verification easier for everyone involved.
If you’re working through any of this, connect with Ronda on LinkedIn.
