Episode 117 | The AI Hat Podcast
Hosted by Mike Allton
Right now, one of your top performers just uploaded a signed NDA into Claude to summarize contract terms. Your partnership director fed proprietary pipeline data into Perplexity to research a competitor. A sales manager pasted customer payment details into ChatGPT to draft a follow-up email.
They aren't being reckless. They're being resourceful. Because 70% of their day is admin drag, and AI is the only tool fast enough to keep them from drowning.
But every single one of those actions just created a compliance time bomb you won't discover until the audit - or the breach.
This is the Shadow AI crisis every Revenue Leader faces in 2026. You're stuck between two nightmares: your team using ungoverned AI and falling behind competitors who are moving faster. The question isn't whether your people are using AI. They already are. The question is whether you're managing the risk or pretending it doesn't exist.
Financial services learned this the hard way. Banks tried to ban ChatGPT. Credit unions blocked access to Claude. It didn't work. Teams found workarounds. They used personal accounts. They accessed tools on mobile devices. The Shadow AI problem got worse, not better.
So the smartest institutions pivoted. They built governance frameworks based on a principle we call "Amnesty and Orchestration" - you can't ban AI, you have to sanction it. You surface what's already happening without triggering panic. You move teams from risky workarounds to compliant, sanctioned systems. And you do it without crushing the velocity that's keeping your pipeline alive.
What Financial Services Learned That Your Revenue Team Needs to Know
Dr. Angela Murphy - known across FinTech as Payments Elsa - has been in the trenches of this transformation. She's a PhD strategist who advises banks, credit unions, and financial institutions navigating the GENESIS Executive Order, faster payments infrastructure modernization, and the brutal regulatory scrutiny that comes with being first to face the AI governance reckoning.
Angela is a graduate of the PayTech Women Executive Academy, the Money20/20 RiseUp program, and was named the 2024 Emerging Trendsetter by PayTech Women. She's a Founding Fellow of Coruzant Technologies and guest lectures at UC Berkeley and the University of Missouri's Bloch School of Business. When financial leaders need to understand how AI impacts underwriting, payment disputes, compliance workflows, and customer data protection, Angela is who they call.
In this episode, she reveals the exact playbook financial institutions used to move from chaos to control - and why the lessons from FinTech are the canary in the coal mine for every B2B revenue organization.
The Three-Step Governance Audit You Need to Run This Quarter
Before Legal runs it for you, Angela walks through the Monday morning checklist every Revenue Leader should complete:
Step 1: Do you have an AI policy?
Not a vague "use responsibly" statement. A real policy that defines what tools are sanctioned, what data can be processed, and who owns governance within your organization. If you don't have one, who needs to own it? IT? Revenue? RevOps? Decide this week.
Step 2: What AI is already in your tech stack?
You likely already have AI capabilities embedded in Salesforce, HubSpot, Gong, or Microsoft 365. Are your teams trained on them? Do they know how to use Copilot without uploading PII? Can they leverage Einstein without creating data leakage? Your reps are using open-source tools because they don't know the sanctioned options exist.
Step 3: Where are your pain points?
Shadow AI emerges where admin drag is highest. Payment processing teams at banks were uploading spreadsheets into ChatGPT because manual transaction lookups were killing their capacity. Where is your team hemorrhaging time? That's where unsanctioned AI is happening right now.
Angela shares real stories from financial institutions that discovered compliance nightmares - HR professionals uploading benefits data with health information into Perplexity, operations teams keying transaction records into ChatGPT, underwriters using AI for credit decisions without explainability frameworks. These weren't malicious actors. They were overworked professionals desperate for efficiency.
The difference between a governance disaster and a competitive advantage is whether you architect the solution before the breach forces your hand.
Why Explainability Isn't Compliance Theater - It's Revenue Protection
One of the most powerful concepts Angela unpacks is explainable AI - the ability to defend why your AI made a specific recommendation.
When your AI agent suggests discounting a deal by 15%, can you explain to your CFO why? When it deprioritizes a high-value prospect, can you audit the logic before you lose the relationship? When it flags a payment as fraudulent, can you defend that decision to the customer - and to regulators?
Financial services operates under scrutiny that's coming to every industry. Angela shares the FICO score example: credit scoring systems built in the 1950s by two brothers in Minnesota, at a time when women couldn't open bank accounts without a man's permission. If you tell your AI to tie credit decisions to FICO scores without accounting for historical bias, you're not just perpetuating inequity - you're limiting your addressable market and opening yourself to legal liability.
Ethical AI isn't a "nice to have." It's revenue protection. One biased recommendation can tank a partnership, torch a customer relationship, or trigger a PR crisis that costs you more than any efficiency gain AI ever delivered.
How Ethical AI Unlocks Hidden Revenue Channels
Here's where Angela flips the script. Governance isn't just about risk mitigation. When you use AI ethically - with clean data, proper parameters, and explainable frameworks - it becomes a revenue growth engine.
Feed your AI five years of historical sales data. Ask it where deals consistently fall apart. Where outlier successes happened. What patterns emerge in your longest sales cycles versus your fastest closes.
The AI might reveal that your 30-year legacy product is bleeding deals every quarter - time to sunset it or reframe the positioning. It might show that an "outlier" win in a vertical you've ignored represents an untapped revenue channel worth building a team around. It might surface biases in your sales process - are you systematically undervaluing prospects from certain industries or company sizes based on assumptions that don't hold up under analysis?
This is Revenue Architecture in action. Workflow first, tools second, governance always.
The Regulatory Hammer Is Coming - Financial Services Just Got Hit First
Angela breaks down the GENESIS Executive Order and why it matters to revenue teams outside FinTech. While the executive order focuses on AI infrastructure and scientific discovery, it's laying the groundwork for cascading regulation across every industry. Financial services regulatory bodies - the OCC, FDIC, NCUA - are watching closely. What becomes standard practice in banking will influence requirements for healthcare, insurance, professional services, and B2B SaaS.
The question isn't whether AI governance mandates are coming to your industry. It's whether you'll be ready when they arrive.
Your Monday Morning Action Plan
If you're a VP of Sales, CRO, or RevOps Director realizing you have a Shadow AI problem, here's what Angela recommends you do immediately:
- Surface what's already happening - Have the conversation with your team. Create amnesty. Make it safe to admit they're using ChatGPT for research or Claude for summarization. You can't govern what you can't see.
- Audit your sanctioned tools - What AI is already available in your existing stack? Get training materials from your vendors. Microsoft offers Copilot resources. Salesforce has Einstein enablement. Use what you're already paying for.
- Identify your biggest pain points - Where is admin drag crushing your team's capacity? That's where you need AI governance frameworks most urgently.
- Build your starter policies - You don't need perfection. You need three things: a usage policy (what's sanctioned), a data classification system (what can be processed where), and an education mandate (regular training requirements).
- Bring in an expert - Whether it's a workshop on prompt engineering, a governance audit, or strategic advisory, don't try to architect this alone. The cost of getting it wrong is exponentially higher than the investment in getting it right.
Resources & Next Steps
Download the Executive Guide to Shadow AI
Get the complete framework for turning security risk into competitive advantage through proper governance: theaihat.com/shadow-ai
Connect with Dr. Angela Murphy
LinkedIn: linkedin.com/in/paymentselsa
Instagram & TikTok: @PaymentsElsa
YouTube: Payments Elsa
Need AI Governance Training for Your Team?
Mike delivers custom Revenue Architecture workshops for B2B organizations. Learn how to architect sanctioned AI systems that protect your pipeline, your customers, and your board credibility. Contact: theaihat.com
Subscribe to AI for Revenue Leaders: The AI Hat Podcast
Stop being a Pilot Purgatory statistic. Every week, we deconstruct how Revenue Leaders are building governance frameworks, digital crews, and operational systems that eliminate admin drag and reclaim their team's capacity for high-value work.
Full Episode Transcript
(Lightly Edited)
Why Your Sales Team's Shadow AI Is a Lawsuit Waiting to Happen - A Fintech CRO's Governance Playbook
[00:00:00] Angela Murphy: When you use AI ethically within your organization, it increases your revenue potential because when you give it the proper data set, and then when you set the parameters correctly, you could feed an AI tool, historical data on your sales within your organization for the last five years. And then you can ask it, with correctly engineered prompts, questions about the sales cycles, where sales deals fell apart, where outlier sales deals were successful, those types of things.
[00:00:40] And it can run scenarios and analyses on those to maybe show you that the outlier might be a new revenue channel for your team. It could show you that you're constantly losing deals on a side of the business that you've had for 30 years, and maybe you need to either reframe that product or sunset it because it's just not the next phase for your company.
[00:01:02] So I think that from a revenue perspective, it can show you where the gaps are, and then also it might show you where there were some biases in your sales cycle or in the way that you approached your product, and it can help you see the potential for new revenue channels for your teams.
[00:01:24] Mike Allton: Mike's got The Hat, the AI kind. Unlocking power in business, you'll find. Cut through the hype, explore the tech. Every week, new ideas to check.
[00:01:46] Allton leads us on a digital quest. AI secrets, he'll always invest. Tune in tight, don't miss a beat. Knowledge, groove, keep you on your feet. Listen up! To the AI flow. Business growth, here we go. Cuttin' edge insights, that's the plan. With The AI Hat, you'll understand.
[00:02:32] Greetings program. Welcome to the AI Hat Podcast, the resource for revenue teams who are ready to stop being the bottleneck in their own business. Each week we architect the simple AI operational systems that eliminate admin drag, and reclaim your time, freeing you and your team to focus on the high value work that truly scales your revenue.
[00:02:51] I'm your host, Mike Allton. Subscribe and let's build your digital team together.
[00:02:56] Now, right now, somewhere in your organization, a sales rep just pasted a customer's payment details into ChatGPT to draft a follow up email. A partnership manager, upload a signed NDA to claw to summarize terms. And a marketing director fed proprietary pipeline data into perplexity to research competitive positioning.
[00:03:16] They're not malicious, they're desperate, because 70% of their day is admin drag, and AI is the only tool fast enough to keep them above water, but every one of those actions just created a compliance time bomb you won't discover until the audit or the breach. Financial services learned this the hard way.
[00:03:35] They thought they could ban AI. They couldn't. So they built something smarter. Governance frameworks that let teams move faster and sleep at night. Today's guest architected exactly that transformation, not in theory, but in the trenches of real time payments, regulatory scrutiny, and executive pressure to show ROI yesterday.
[00:03:54] Angela Murphy, known across FinTech as Payments Elsa, is a PhD strategist who advises banks, credit unions and financial institutions on AI integration governance and the brutal realities of faster payments infrastructure. She's a graduate of the Paycheck Women Executive Academy, the Money 2020 RiseUP program, and was named the 2024 Emerging Trendsetter by PayTech Women.
[00:04:16] She's a founding fellow of Coruzant Technologies and guest lectures at UC Berkeley, and the University of Missouri's Block School of Business. Her insights have been featured on the Treasury Whisperer, the Payments Podium, and Wrestling Payments, and she spends her days helping financial leaders navigate the Genesis Executive Order, explainable AI mandates, and the existential question Every revenue leader faces, how do you govern AI without killing velocity?
[00:04:43] Hey, Angela, welcome to the show.
[00:04:46] Angela Murphy: Hi there, Mike. Thanks so much for having me. I'm excited to be here.
[00:04:51] Mike Allton: My pleasure. I can't wait to dive into this with you. I've got so many questions. I wanna start off, Angela, with your work in financial institutions and they're under intense regulatory scrutiny around AI.
[00:05:02] So when you walk into a bank or a credit union, start asking about their AI usage, what's the first thing you usually discover and how often does leadership already know about it?
[00:05:12] Angela Murphy: This is one of the most shocking things that has come up for me as an expert in artificial intelligence and financial services, and particularly in the last year to 18 months with the rise of Chat GPT and kind of the public realm.
[00:05:28] When I speak to leadership and financial institutions or people who work directly with their leadership at conferences, the first thing that they say is, we're never gonna use it. And I can't believe it. Right. We're, I'm sure there are several people who are listening to this right now scratching their heads saying, how can you say that? I use it every day. And I can tell you from experience, that's one of the first things that they say to me. Or when I'm giving workshops or lectures or webinars, I'll ask for, you know, a show of hands. How many of you have used chat GPT or some other form of AI tool in the last six months? And at the beginning of last year in a room of maybe 50 people, I would see three to five hands.
[00:06:13] Even by the end of the year. It still wasn't half the room. It wasn't quite there yet. There's a lot of. Fear, I think in financial services about artificial intelligence and how it's used because of the way that their data is guarded and governed by regulatory frameworks. And I think it's also just a bit of confusion from the general population of, you know, is AI dangerous?
[00:06:39] Is AI gonna steal my job? What does this look like for me? And in a fairly manual environment like the back office for a banker or credit union. All of those questions are top of mind. And when you're looking at the board or the leadership, because they're trying to protect their institution first and de-risk any technology that they would engage with, they're gonna say, Nope, we're not gonna do it.
[00:07:02] And so I've seen several leaders actually put a policy out that says, we are never gonna use it, we're not gonna touch it. Which I think is an error.
[00:07:10] Mike Allton: Oh, it's, it's definitely an error. Their teams are definitely using AI without their knowledge. They're just making them scared. Uh, we saw that with Samsung who tried to ban ChatGPT couple years ago, and it just doesn't work.
[00:07:23] So let's, let's talk about that, that shadow AI problem. Initially and, and specifically, and I'm wondering if you could share some stories or some insights from, from this industry where they've got a governance gap and that might've turned into a disaster 'cause those team members are already using AI on the on their own.
[00:07:41] Angela Murphy: Yes, I have definitely seen it in one of a few ways. So in financial services, obviously you have HR professionals who might be putting benefits documents into a Perplexity, a Claude, or a Chat GBT to do benefits summaries, which can be a little bit dangerous because some of that has to do with like health data, but also in terms of the way that financial institutions operate, I think sometimes that there's a misconception that they have strong technology, and for my banking, credit union and financial services friends that are listening to this, we all know where the skeletons are. So don't come for me in the comments. All right? There's a lot of manual processes that underpin how money moves in the United States, and we just have to call a spade a spade.
[00:08:29] And with that back office operations challenge comes the administrative overload of. Keying information from one spreadsheet into another that then goes into a ledger or a system of record, and then you have to go back and find data on a particular transaction. And so there are some cases where people have uploaded a spreadsheet into a Chat GPT, and then asked it or queried it for a specific transaction or something about a transaction or data associated with something that's going on within that spreadsheet. And that can be really dangerous because a lot of the information that financial institutions touch is what we call PII, Personal Identifiable Information.
[00:09:16] So that's account and routing numbers, name, address, et cetera. And you don't want that information out there. But when you have people who have too much to do, essentially in terms of manually processing financial data. They're gonna be looking for ways to shortcut that. And this is especially true, I think, as we are looking at attrition in financial services, we have a lot of people with decades of expertise that are retiring, that are not being backfilled.
[00:09:50] And one of the things that I wanna talk about that is pretty well known as public information is that there was a very high profile bank, CEO, that has come out in the last couple of months and said, we're freezing hiring for three years and we're just gonna use AI to do those jobs instead. I just find that to be a massive gap in the understanding of not only the governance, but the experience that is necessary to have the correct governance for how to use artificial intelligence for these highly specialized roles.
[00:10:29] Mike Allton: Yeah, that's exactly right. Like you mentioned, you know, someone using like a Chat GPT to help them find specific transactions and data in a spreadsheet. Sounds like a great idea at the surface, but you and I know that there's huge flaws in that approach because of the PII sharing and so on. There's ways that you can use AI for that task, but it requires access to paid tools.
[00:10:54] It requires training, uh, and oversight so that the employee knows what they can and can't do in the right way. For the right type of task. So for revenue leaders that are listening you know, maybe they're not in financial services, why do you think they should care about what's happening in FinTech right now?
[00:11:14] What's, what's the canary in the coal mine lesson here for them?
[00:11:17] Angela Murphy: I think for revenue leaders, at the end of the day, it. Do it is in your job title, right? Revenue. That means money and how money affects your organization. Well, how money affects your organization is inherently controlled by financial technology, FinTech, and technology is what is a part of your business relationships. It's a part of how employees are paid within your organization. It's inherent in how people pay you for your goods and services and how you run your business. And so when you look at FinTech or financial technology. As an umbrella term, you're talking about banking, you're talking about invoicing, you're talking about payments, and quite frankly, if anybody has been watching the news since last summer, there is some emerging technology in terms of how money is going to move in the US differently because there is emerging payments infrastructure.
[00:12:21] It started in 2017 with instant payments from the clearing house. The Fed had an answer two years ago with the Fed Now Service, and those are both instant payments rails that have FinTech overlays to provide access to that capability. And then in the last year, we've seen with the passing of the Genius Act that the US is now getting involved in something called stable coin, which is another piece of instant payments infrastructure.
[00:12:51] So as a revenue leader, the technology that underpins how dollar goes from here to here is critical to you, to your role and to your business.
[00:13:03] Mike Allton: That makes a lot of sense. Now, one of the things I mentioned at the outset was this Genesis executive order. I'd love for you to explain that to those listening.
[00:13:12] Frankly, I'd love for you to explain it to me for those who haven't tracked it. Maybe break down, you know, what, what it requires from, uh, you know, FinTech institutions. What that means for, for revenue teams who are trying to think about AI governance in 2026.
[00:13:27] Angela Murphy: So this was something that was really interesting to me because the Genesis Executive Order, and I do wanna be really clear, there are two pieces of legislation that legislation, I say with quotes that start with a G, that have to do with things in FinTech, and one of them is the Genius Act, which was a congressional act, was passed through Congress.
[00:13:48] It was signed. Into law by the current president last summer, and then there is the Genesis Executive Order for those of us who might have failed high school civics. Those two things have different implications in terms of the weight that they hold and how they can be carried out. So keep that in mind and kind of check on that on your own.
[00:14:09] But those two things are different and operate differently within our kind of legislative and executive branches of government. But the Genesis Executive Order specifically came out the week of Thanksgiving, so nobody really talked about it, and I thought that was fairly interesting. But it essentially was an executive order that was talking about the framing of infrastructure for artificial intelligence capabilities within the US.
[00:14:37] And infrastructure is mentioned a few times in the executive order about how we need to build out the capabilities for the United States to be a leader in artificial intelligence infrastructure and how we use that. But one of the interesting things is that it also tied it to economic growth. So when you see legislation or executive orders that says anything about economy, economic growth, those types of things tend to have cascading implications for financial services. So while this executive order seems on the surface to focus primarily on infrastructure and scientific discovery, it's laying the groundwork for the potential of additional legislation and regulation that could affect financial services.
[00:15:30] When I see things like this, I call it something along the lines of suggestulation. So an executive order can't necessarily tell a bank exactly what to do but. The OCC, FDIC, the NCUA, these entities that create the regulatory frameworks that financial institutions do have to listen to are paying key attention to executive orders like this one, because it's a step in a journey to where the use of artificial intelligence will become part of the bedrock of how a lot of institutions are gonna be required to do business and to engage with their account holders.
[00:16:12] Mike Allton: Thank you for that. I hope folks understand that and definitely take a few moments to do some additional research. Ask Perplexity about this particular executive order and, and ask how it applies to your organization. Now, one of the frameworks that we talk about a lot for organizations are, are amnesty and orchestration.
[00:16:30] This idea that you can't, or at least shouldn't ban AI, you've got to govern it. How do you actually have, in your experience that conversation with a leadership team that to your point earlier, they're, they're terrified of both risk and falling behind?
[00:16:45] Angela Murphy: One of the things that I try to talk to. Financial services leaders about is that fear hasn't really ever served anyone when it comes to moving money.
[00:16:57] We, in our industry have always had to look to innovate because of the behaviors and the needs of our account holders. So. Obviously there is a lot of fear around artificial intelligence and what happens if we break it or we do the wrong thing, and I like to address it in terms of taking apart the problem in bite-sized pieces.
[00:17:26] What I ask them to do is to look within their institutions and to see where the highest pain points are. So where are the most manual processes within your institution? Where's the most strain? What problems or processes within your institution are costing the most time people and money? So resources.
[00:17:50] And then once they've kind of gone through that exercise from a strategic perspective, it becomes pretty clear the between one and four buckets, if you will, where technology could probably solve a problem in one of those buckets. And so it's more about looking for the low hanging fruit within those institutions as to what problems technology does need to solve, and then going through the due diligence process like FIs normally need to do of evaluating technology partners that can help solve for those problems. So there are a lot of great fintechs out there that are focused on artificial intelligence in financial services specifically, and they have guardrails and they have boxes that keep data safe. And they have the ability to make sure that if an institution works with them, that that data set stays within the four walls of that institution instead of being accessible by outside parties.
[00:18:56] And so I really tried to demystify it and tie it to existing processes that financial institutions are already aware of since they already probably work with a third party service provider. Or sometimes you'll see it referred to as TPSP.
[00:19:13] Mike Allton: Got it. Folks we're talking with Dr. Angela Murphy about how organizations, revenue teams, can address the unmonitored and unrestrained use of AI and get some lessons from FinTech. And I've got several more questions you aren't gonna wanna miss, but first, a quick resource I have for you on Shadow AI.
[00:19:29] Right now, your best people are drowning. They're buried in data entry, meeting prep, and internal reporting. The admin drag that kills momentum. To keep their heads above water. They're turning to AI tools you haven't approved. They aren't trying to be reckless. They are trying to be efficient. They are trying to hit the goals you set for them, but this Shadow AI creates massive blind spots in your data security.
[00:19:57] You can't afford to ignore it, but you also can't afford to ban it. You need to architect it. I've created the executive guide to shadow AI to help you bridge this gap. It's a blueprint for turning a security risk into a competitive advantage through proper governance. Don't fight the shadow, sanction it.
[00:20:20] Download the free guide today theaihat.com/shadow-ai. That's theaihat.com/shadow-ai.
[00:20:33] So once you've, you've talked to leaders, maybe you talked to people on the team, maybe done some surveys or something, and you've surfaced what people are actually using. What would the orchestration part look like to you?
[00:20:43] What are some initial policies or systems that you would help organizations put in place to start?
[00:20:49] Angela Murphy: So, some of this is going to seem a little bit obvious at the outset, but in my work in artificial intelligence and financial services, I like to start with just the most basic level of everything, and that's to one, make sure that your teams know that it's not okay to just upload spreadsheets into Perplexity or Claude or any of these other things. So that seems like it should be just table stakes and the basics, et cetera. But sometimes it's not and you don't know what your teams don't know.
[00:21:32] So in terms of orchestration, I tell them to get a level set. So figure out what tools your teams might be using and what tools are available to you in your existing technology stack. And then to say, okay, if it's already something that's within our existing technology stack, this is okay. Once you've gone through the evaluation process to make sure that it's secure and that the data within that solution is secure.
[00:22:05] Then I would say that there needs to be an education part of this, and that to me is something that I am quite passionate about.
[00:22:15] Not everyone is an AI expert, and if we're in financial services, it's highly unlikely that we are, especially if we are leading a large organization or even if we're leading a community financial institution. So part of the policy is probably gonna be about regular education requirements for artificial intelligence and best practices. Also teaching people how to use the tools that are already available to them.
[00:22:42] And then the third thing would be to model your data policy and your data governance policy off of what you've already established as best practices. Most banks and credit unions already have a lot of security policies or policies that they have to be able to turn over.
[00:23:03] To the regulators to prove that they're not bad actors. So it's about looking within the existing structures of the policies you already have in place and then saying, okay, what pieces can we take from here that could be at least a start for a policy about how we're gonna govern AI usage within our institution?
[00:23:23] It does not have to be perfect at the outset, but you do need to start, and if you haven't started already, you need to start now.
[00:23:32] Mike Allton: Got it. So we're going to give them some training. We're gonna give them some policies. They're starting to use AI in certain ways. I've seen you talk about this concept of explainability where you say AI that you can defend.
[00:23:44] What does that actually mean? In practice? What, like, be an example maybe of a decision that, that AI might make in a revenue context that you have to be able to explain that and to whom?
[00:23:55] Angela Murphy: Exactly. This is one of my favorite things to talk about because I don't think that it gets enough airtime is explainability and it ties very closely into ethical ai.
[00:24:06] When we make decisions as humans, we have to be able to defend why we made that decision. So if I go to Costco and then I go to Hy-Vee in that order, but my husband would go to Hy-Vee and then to Costco in that order. We've both made it a decision. It's not that either one of us made a bad decision, we just made a different decision, and we have to be able to tell each other why we did that, right?
[00:24:32] It's just the same basic principle of human reasoning. If you go this way and I go that way, why did I go this way and why did you go that way? And explainability is really important because it's a way of de-risking how you engage with artificial intelligence. We have already seen that there have been some interesting lawsuits that have come out about data usage, about AI usage, about whether or not the AI was making the proper decisions based on the information that it was given.
[00:25:07] And this is especially key in financial services because as I tell people quite frequently. Your artificial intelligence tool is only as good as the information that you give it and the parameters that you set around that information. So garbage in, garbage out, right? And with PII with financial data specifically, look at underwriting.
[00:25:34] If you're going to give a data set of underwriting materials to an artificial intelligence tool and say, please make credit decisions for these 10 files. You have to be able to give it the rules within your institution as to what is credit worthy, what might be credit worthy with a little bit of assistance and what is not credit worthy at all.
[00:25:58] It's the same thing that a human would do in that scenario. A credit decisioning person in the back office at a bank. And in both of those instances, you have to be able as a bank to say why you denied these applications, whether a person did it or whether an AI did it. But let's take that one step further because in credit decisioning in particular, a lot of what gets relied on as to whether or not you're a good person with good credit in the US is a FICO score.
[00:26:30] Well. If you don't know your FICO score, please go figure that out right now, because everybody should know what that is. But
[00:26:36] Mike Allton: after the podcast,
[00:26:38] Angela Murphy: after the podcast, but go do it. If you don't know what your FICO score is, go find it. But there's an issue with that historically because for those of us who don't know, FICO scores weren't created until the 1950s and they were created by two brothers out of Minnesota.
[00:26:54] And that's kind of an issue because I, I, people can't see me on this podcast, but I am a woman, and at that time when FICO scores were created, I wasn't allowed to have a bank account without a man on the bank account. So that score didn't really apply to me because I didn't have financial autonomy. So if you're in this case looking at credit decisioning and using artificial intelligence, but you tell it, to tie it to a FICO score, historically that's problematic. And then you're basically telling the AI to follow rules that aren't necessarily going to expand your credit profile. And to make sure that you're making strong credit decisions within your institution.
[00:27:38] You have to create rules and parameters that allow it to take in other pieces of data so that it can make better decisions than what we've been making historically within our institutions.
[00:27:51] Mike Allton: That is a huge point, and it's kind of leading into not just the ethics of finance, but the ethics of AI. I know AI has a lot of bias.
[00:27:59] I know that's something you've been very vocal about. Uh, and I have as well, we've had terrific guests like Dr. Nici Sweeney on the show just to talk about AI and ethics and bias. But I'd love for you to just share what you think for a second. How does that connect to revenue outcomes? You know, why should a VP of sales care about whether their AI recommendations are biased?
[00:28:20] Angela Murphy: Well, one, you don't wanna get sued. That's a big one. But we have seen this. It is not just that the lawsuits around artificial intelligence are, why are you using my data? It's copyrighted. But it, some of these lawsuits have also been about people not being given lines of credit. Because the AI made the wrong decision, it made the wrong assumption.
[00:28:44] And so when you talk about ethical AI from a revenue leader perspective, you want to make sure that your organization is always above board. And also when you use AI ethically within your organization, it increases your revenue potential because when you give it the proper data set, and then when you set the parameters correctly, understanding that there are institutional biases and that there can be coding biases.
[00:29:14] 'cause once again, it's not just about the data that goes in, it's about the way that the AI is coded, right. It's going to be code it in a specific way to do a specific task in a specific way until you tell it to do something differently. And the potential here is that you could feed an AI tool, historical data on your sales within your organization for the last five years, and then you can ask it, with correctly engineered prompts, questions about the sales cycles, the decisions that were made where sales deals fell apart. Where outlier sales deals were successful, those types of things. And it can run scenarios and analyses on those to maybe show you that the outlier might be a new revenue channel for your team. It could show you that you're constantly losing deals on a side of the business that you've had for 30 years, and maybe you need to either reframe that product or sunset it because it's just not the next phase for your company.
[00:30:20] So I think that, from a revenue perspective, it can show you where the gaps are, and then also it might show you where there were some biases in your sales cycle or in the way that you approached your product, and it can help you see the potential for new revenue channels and for new training opportunities for your team.
[00:30:45] Mike Allton: I love that point because it's so true. The, the correct use of AI with correct prompts. Can absolutely help you identify your own biases and your own biases in your team or your data, your past marketing you know, that, would steer you away from thinking that the majority of your clients are affluent or something along those lines, and allow you to make proper decisions about who you're talking to, how you're talking to them.
[00:31:08] Terrific points. Thank you for sharing those. I think folks really you need to go back and listen to that one more time and really just let that sit with you and let you think about how you're going to apply this in your organization. Just a couple more questions for you. I, I wanna know, what do you think is the biggest misconception maybe that revenue leaders have about AI governance?
[00:31:28] What do they think that it requires maybe that it actually doesn't, or maybe something that they assume is optional that's absolutely not.
[00:31:36] Angela Murphy: One of the things that I think is a common misconception is that. Well, like we've talked about today, that you can ban it. You can't. But that you have to know every single thing about every single AI and every single way that your team might use it in order to create the perfect policy so nothing bad ever happens to you. It's a bit of a chicken little effect. I see with revenue leaders where they have this feeling that if they don't understand all things for all people and all technologies, that one minor issue could be the string that unravels the entire blanket. And if you are going about this process with due diligence, there's gonna be so many checks in a row that's gonna prevent it from being a governance disaster, that you don't have to have that level of concern about it.
[00:32:35] There are trusted voices out there. There are trusted companies out there. There are case studies out there that you can look to, that you can do research on and model your best practices off of that just like you've done in other sides of your organization.
[00:32:51] And the thing that I see the most that revenue leaders think is optional, that absolutely isn't is the education piece. And for anybody who's heard me speak on these topics before I came out of academia, I think that education is our number one best tool to ensure that our teams are optimized and when you look at artificial intelligence and the need for it within our organizations, I think sometimes where organizations fall flat is they just say, okay, we're just gonna use copilot.
[00:33:27] Be free, go forward, it's gonna be great. And then don't bring in some type of expert to sit down with their teams and talk about the strategies that they can use individuals within the tools to show them the options that they have within prompt engineering to show them how they might be able to "agentify" the tool that they're using, and I don't think that enough revenue leaders bring in the experts to talk about it. I think that there's a misconception that you just flip a switch and it's gonna be okay, and in some cases it might be all right, but if you're going to bring artificial intelligence into your organization.
[00:34:14] The best thing that you can do is to bring in an expert to speak with your leadership about the AI strategy within your organization, and then you go through the evaluation process and you select your tools, and then you find experts on those tools. Then you bring in the experts to talk to the individuals that are on these teams that are gonna be engaging with them.
[00:34:34] I know some of the most popular workshops, webinars, talks that I have given in the last year, and my colleagues as well have been just about prompt engineering and helping people understand how they can teach an AI to think like them and to problem solve like them. And if you're bringing someone in to talk to your team specifically about prompt engineering, it's also a great way to de-risk your organization because that person can show them, Hey, don't upload the document, but look at the document and the questions that you have, and then this is how you can design questions to ask the AI about the situation. To help you find the answer that you're looking for, and so I cannot stress enough that bringing in the experts is something that's not optional.
[00:35:26] Mike Allton: Oh, that is, that is so true. I teach a five week cohort. Uh, I've been doing it, uh, mostly for this medical manufacturing company.
[00:35:34] Out of, you have this massive company, they've got, you know, tens of thousands of employees. Thousands of marketers, so they're doing these little cohorts. Right. And I've been so proud of these marketers, product marketers and copywriters and so on who are going through this cohort and they're learning Copilot, Microsoft Copilot, to your point, many of them, this is the first time they've ever opened it.
[00:35:51] But I'm equally proud of the company for making that investment, right? And helping these people. And not just saying, to your point, here's a Copilot license. Go have fun. No, here's how we're going to use it. Here's how we're gonna use the data responsibly. Here's how we're not gonna put PII into Microsoft Copilot publicly.
[00:36:08] They've got a private GPT for that sort of thing. That's fantastic. And to your earlier point, I could not agree more about how revenue leaders in particular don't need to know everything there is to know about AI. This is a story I haven't really told on this show before, but you know, my first business, my first blog was The Social Media Hat, and that's because.
[00:36:28] In an earlier career, I was working for an IT company and I watched the owner of that IT company wear every hat in the organization. He was the CEO, he was the HR manager, he was the sales manager. He was the implementer of, of servers or whatnot. He was the janitor, the plumber, the painter, everything.
[00:36:47] Right? Because it was a small business, he had to wear all those hats. And so my thought process with The Social Media Hat was when one more hat is one too many, let me wear the social media hat for you, for your organization, and I carry that forward with The AI Hat. What I do here is I tell people the things that they do need to know, and I filter out for you all the stuff that you don't need to know.
[00:37:09] There's so many AI foundational models out there I don't even talk about. We only talk about a fraction of them. When we talk about Claude, Gemini, Copilot, there's thousands more around the world. You don't need to know that. You don't need to know what the latest feature was. You just need to know how do I make this work for my organization?
[00:37:30] So my last question for you, if someone's listening to this, they're realizing, eh, they probably do have a Shadow AI problem in their team. They know their reps are using Chat GPT, their marketers are using Jasper, their ops team is using something else, what's the first thing that they should do Monday morning?
[00:37:47] What's the audit or the conversation or the policy that you think they should start with?
[00:37:52] Angela Murphy: I first of all think that you need to figure out if you even have a, have an AI policy at your organization. So that's question one. Do we have a policy about AI usage within our organization? That's gonna tell you a lot, very quickly about the next steps, which is if we don't have an AI policy, we need to create one and decide where that sits, who owns that within the organization?
[00:38:17] Because it could be IT, it could be revenue, it could be somebody else, but you will need to have one because it does touch your teams. So one, do we have an AI policy? And if we don't, who needs to own it and how can we get this set into motion this week?
[00:38:39] And then two, ask what AI tools are currently available to our teams within our current tech stack?
[00:38:48] So not open source, but within our organization, are there AI tools that are built into our current technology stack? And when you figure out if there are or aren't, asking what those are, who the experts on your teams might be about them, or maybe talking to your rep. So let's say it is Co-pilot, can you talk to your rep and use them as a resource?
[00:39:16] Maybe have them send over some training materials that are pre-canned to the teams that could use Copilot, that could optimize their usage of Copilot.
[00:39:26] And then the third thing is. Where in our organization could we be utilizing technology, but specifically AI to solve for pain points within our organization?
[00:39:41] Because that's really what it is, is a tool in a toolkit. Just like you would hire an intern to do data entry over the summer like I did at a bank eons ago when I started this mess. AI is going to be something that is a part of your ecosystem, and you already have the team members in place to be able to add the parameters, to add the human in the loop, to add the institutional expertise into that tool and that ecosystem to make it work for you.
[00:40:17] So the final thing that I would say is that, if you get pushback as you're starting to ask these questions within your organization, particularly around, well, I don't want a robot to steal my job. The thing is, the artificial intelligence is only as good as the humans that are around it, and it's not going to replace any person that has a level of expertise that is contributing to the conversation.
[00:40:44] Mike Allton: That is fantastic advice. You have been such a wealth of knowledge. I wanna thank you for coming on this show, but I know folks listening their minds are on fire with questions. They're gonna wanna reach out to you. Where should they go to get more information or to touch base?
[00:40:59] Angela Murphy: I'm pretty easy to find online, so you can find me on LinkedIn.
[00:41:04] You can find me on, uh, other social media, YouTube as well at Payments Elsa. So if you wanna find me on LinkedIn and follow me there, it's just linkedin.com/payments elsa, I am the top search result. It'll say Angela Murphy, PhD, and then I'm also on the clock app. And, uh, and Instagram as well for my informational videos.
[00:41:30] Mike Allton: Fantastic. We will have all of Dr. Murphy, a k a Payment Elsa, all of her links, uh, in the show notes below. So thank all of you for listening. That's all the time we've got for today. If you haven't yet talked to your team about how they're really using ai, shadow ai. It's obviously time. You can learn what Shadow AI is in more detail, why you should lean into it, and how to proceed for free at theaihat.com/shadow-ai.
[00:41:56] Check it out and I'll see you in our next episode. Welcome to the grid.
[00:42:00] That's a wrap on another episode of The AI Hat Podcast. I hope you've gained some valuable insights and actionable strategies to boost your business with AI. If you enjoyed this show, please subscribe, leave a review, and share it with your network. And don't forget to visit the ai hat.com for more resources and to connect with me.
[00:42:19] Until next time, keep innovating and remember, the future of business is human and it's powered by AI.
