In this episode of the Sunny Ray Show, Sunny Ray talks with Tim Atzinger, co-founder and president of Ovation CXM, a customer journey orchestration platform for banks, payment providers, and fintechs. Tim explains how his 30 year career at Visa, First Data, Blackhawk, and Monetize showed him how fragmented commercial banking go-to-market processes are, which inspired him and his co-founder to build an orchestration layer connecting internal systems, teams, and ecosystem partners. The conversation moves into Ovation CXM's newest focus, the Agentic Transaction Control Layer, or ATCL, which lets banks govern, permission, and audit the AI agents that commercial clients are already using to handle treasury tasks and financial workflows. Tim shares survey data showing most financial officers are ready to bring AI agents to their banks, discusses open questions around liability and regulation, and explains why banks should buy rather than build this capability. He closes with a bold prediction that most banking transactions will become agentic within the next few years, reshaping how banks compete for commercial customers.
Ovation CXM's Tim Atzinger on why banks need a governed front door before AI agents start banking for customers.
What is Ovation CXM and what are you building right now?
We built Ovation CXM to help banks orchestrate customer journeys across their siloed internal systems, teams, and ecosystem partners, and we have been driving that in North America for several years. Our most recent effort is helping commercial bank customers' AI agents access those journeys, since CFOs are increasingly using agents to perform financial close and reporting tasks that require engaging the bank, and banks are woefully unprepared to respond.
You spent 30 years at Visa, First Data, Blackhawk, and Monetize and kept hitting the same wall. What was it?
Financial institutions go to market in an extremely fragmented process. They tend to be functionally organized, and when a customer is trying to accomplish something across those functions, teams, and systems, it can be super disjointed. There was nothing tying those systems and people together so you could get an end to end view of what a customer is trying to accomplish.
What does a customer falling through the cracks actually cost a bank in real numbers?
Pre-activation churn is a huge challenge in commercial banking. With some treasury products, as much as a third of the customers you sign never go live, and 40 to 50 percent of those who do go live have ongoing problems. When we fix this with the platform, banks see a 10 to 15 percent revenue lift, 20 to 25 percent attrition reduction, and 20 to 30 percent cost reduction.
Can you define ATCL for someone who has never heard the term?
ATCL is the Agentic Transaction Control Layer. It is a way of saying I am going to govern who shows up at the front door, know they have permission to do certain things, track what they are doing, report on their activities, and turn off access if needed. It is the equivalent of permissioning human beings into a bank system, but for AI agents tied to those humans.
Where did the 72 percent stat come from, and what worries the other 28 percent?
We surveyed over 500 financial officers across commercial clients in the US, and 72 percent said they are on board with letting an AI agent operate at a bank on their behalf, with over 50 percent already using agents internally. The other 28 percent had concerns around fraud, security, permissions, and audit trails, which is exactly what our ATCL solution helps solve.
If an agent does the wrong thing with someone's money, who is liable, and does the regulatory framework have an answer?
That is a really hard question and there are a lot of unsolved issues in the industry right now. We think the most straightforward way to tie a regulatory framework to agent activity is to match it to the same permissions given to the human user, since agents are essentially automated avatars of a client. We have built governance and audit into ATCL to manage that.
Why would a bank buy the control layer from you instead of building it themselves?
Banks cannot possibly build everything internally, and most IT budget is tied up maintaining existing systems. Why spend time building registration, identity, permissions, execution, audit, and regulatory controls around AI agents when you can get it off the shelf and plug it straight into orchestration capability that already exists at the bank? In financial services it is the application of AI that matters, not the capability alone.
Five years out, does the customer journey still have humans in it, or is it agents talking to agents?
I think it is more the latter. AI starts by helping humans, then AI and humans work hand in hand, and eventually you get to humans overseeing the actions of AI agents because of the leverage it creates. We think most banking transactions will become agentic in the near to medium term, and most banks have a blind spot because they are only testing internal AI use cases.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview