This is a CIO Corner interview (host Juan) with Kyle Mai, Chief Innovation Officer at Esquire Bank, about how businesses can build trust in AI outputs. They cover practical mechanisms such as forming an AI governance committee, prioritizing regulator-safe use cases, and deploying a platform-level "trust layer" to mask sensitive data, reduce hallucinations, and provide audit trails. Kyle describes Esquire Bank—a branchless B2B bank focused on the litigation vertical—that uses AI for hyper-personalized marketing and emphasizes protecting customer data; both discuss the emerging role of AI agents ("agent force") and how chained/managed agents can augment workflows like marketing plus compliance checks. The conversation also cites a widely referenced statistic ("93% of IT leaders do not consider AI outputs completely trustworthy").
Chapters
Every single day as a cio, I'm getting hundreds of emails from different vendors out there offering me some kind of AI solution. And the reality is you don't know if they're going to be there tomorrow, next week or next year. Who are our trusted vendors?
Welcome to CIO Corner. Here's where we will talk to IT leaders about the biggest challenges and opportunities that we all face when it comes to integrating technology into the enterprise. Today I'm joined by Kyle Mai, Chief Innovation Officer at Esquire Bank, a digital first branchless bank that is using AI to drive hyper personalized sales and marketing efforts to reach its customers. Kyle, thank you for joining me today.
Thank you Juan.
You've done a lot of work in terms of creating a framework, guardrails, mechanisms within the organization so that you ultimately build and use responsible and trusted AI. It'd be great for our audience to hear a little bit about what you've done to be able to create those types of mechanisms to support trusted AI.
We're taking things slow and steady and picking use cases that regulators consider safe, our staff and our bank consider safe. The first thing we have is an AI Governance committee. You know, I chair that AI Governance committee and that's critically important to set up a group where they are specifically responsible for understanding the ethical considerations with AI, the people considerations with AI, and the security and compliance related issues as well. You need to have senior executives that are involved and engaged in that group and having meaningful discussions and they're going to be doing things like setting the strategy for AI within your organization, understanding what the use cases are, helping you to prioritize which use cases should be going first.
I absolutely agree. You know, we have formed something similar. We call it the AI Council here and it's got members from our ethics and compliance group, it's got members from Privacy and Legal. To your point, it's a well staffed group with expertise in multiple areas to ultimately make sure that when an idea comes through that we're taking all aspects of building technology into account and create a solution that can generate value for the future.
There's so much changing in the AI space every single day, it's a bit of a headache to really manage. But having that group meet regularly and having the right people from legal and compliance in that group makes the organization feel much more safer with regards to implementing this technology. I've seen a statistic out there that 93% of IT leaders do not consider AI outputs completely trustworthy for work related tasks. Why do you think that's the case and what kind of features do you think need to be built into AI platforms to improve the trustworthy and the accuracy of results?
I'll first start from a product standpoint and the technologies that we make available to customers like you. Kyle. The company has taken the position that with anything we do with AI, first of all, we're going to respect your data. Your data is not our data and your data is not somebody else's model's data. It's your data. The second thing that I would say is that the company's done a really effective job in creating a trust layer across all of our AI solutions. The idea is that whenever you send a request towards our AI, the trust layer will come into action. It will make sure that the data is properly masked. If there's anything there that should not be used for the purpose of generating a response. It will make sure that we address toxicity hallucinations. It will make sure that there's an audit trail so that you can always go back to the source of the data and ensure that the output that comes from our AI is ultimately trusted.
When you talk about your trust layer, to be honest, that's music to our ears because it's just one less thing thing I need to worry about. And knowing that Salesforce is being proactive with regards to communicating what you guys are doing from a security trust, masking data, dealing with hallucinations, perspective, all within one platform. That's gold. It really is. And there are many vendors out there that are going as heavily out into the market and saying that's a key priority.
Curious from your perspective, Kyle, how are you approaching trustworthiness in the data that especially in your industry that is so regulated that data that it's being used to generate AI related outcomes.
We are a disruptor bank. We do not have any branches at all, nationwide presence and we specifically target the litigation vertical and we also have a payment processing arm, but we are straight up B2B. So we specifically have a niche target audience and we use AI to communicate hyper personalized marketing messages to them to ensure that they receive the right content at the right time in a hyper personalized way. As a bank in a highly regulated industry, keeping your customer data safe, secure, trusted is the number one priority for all of our technology initiatives. When you think about how an LLM is actually generated and then the results that you get from an LLM, the LLMs were built based upon generic publicly available information on the Internet, in books, in videos. However, the LLM does not have access to contextual data and information about your organization. And it's critically important to have that in order to increase the accuracy of results and to also ensure that when you're deploying it to your employees and staff members, they're actually getting responses that are meaningful and real to you as opposed to what the LLM was trained on.
That's awesome. And I agree with you. You know, we are really thinking about the way that we deploy AI for our customers in a very simple way. We're. We're thinking that when you have humans, when you have people, and you add AI agents, gen AI capabilities to the work they do, you ultimately can drive more customer success.
You said the magic word, AI agents. AI agents seem to be something everyone is talking about now. How do you see AI agents playing a role in the organization and society as a whole?
You know, we are very active in building a strategy around agents. We think of agents all working together to provide a number of services and capabilities to again complement the work that humans, that people do. We think that there will be a point in which you will have agents communicating with agents. You will have agents that manage other agents customer support. Customer service will benefit immensely from agents. So a lot of opportunity here. And in fact, the company is going down the path of creating what we call agent force, similar to Salesforce. Now we have agent force, where all these agents will be built over time to create enormous capability for our customers in the platforms and the clouds that we support.
I understand the value of agents quite well. I'm a bit of an AI geek. I created an agent for myself and multiple agents, actually, and I use it on a daily basis. And I've always wondered to myself, what would happen if you connected them together and then how could you apply this in a business context? And I could imagine that's going to be a game changer for a lot of organizations when they realize that. So hearing you talk about agent force and potentially having to manage these agents centrally, that's very exciting.
And in your industry, I think service agents will provide just enormous value for your customers. Whether you're in a B2B environment or a B2C environment, those agents will be interacting with one another to provide ultimately better support for the customers that need to interact with the bank.
Yeah, absolutely. I mean, you can create an agent that can help you write a marketing email and then another agent that can check for compliance and make sure that the right words are in there. Then you can have another agent that does something else. So that's really exciting, really, when you start thinking about the kind of capabilities that exist there.
Kyle, thank you so much for joining me today. Please check out the links in the description below and be sure to stay tuned for more episodes of CIO Corner.
