Best Practices for Executive Teams Deploying AI in Financial Services
The rise of large language models and agentic AI is the most transformative technological shift of our lifetimes. But history has shown us how quickly transformation hype can outpace practical value.
Think about the “revolutionary” innovations from the past 20 years. Everyone scrambled to be an early adopter—but too often, they prioritized novelty over purpose. Today, the same story is playing out with AI. It’s like humanity just discovered fire, and everyone’s racing to see what they can ignite without understanding why.
As I meet with CEOs and executive teams from leading mortgage lenders and financial institutions, the conversation has shifted from “What can AI do?” to “How do we deploy AI responsibly, at speed, and with measurable impact?”
Most AI strategies are failing
A recent HFS Research report found that financial institutions plan to increase AI spending by 25% in 2025, with AI now representing 16% of total tech budgets. But bigger budgets aren’t fixing the real problem: most institutions are investing in AI without a clear path to value.
The report also revealed that most AI strategy is being driven by IT, not by business leaders or the C-suite. And the top goal? “Productivity.” Which translates to automation—not transformation.
This approach leads to predictable results:
- 80% of AI projects fail to deliver value
- 42% of companies have abandoned their AI initiatives altogether
I believe these AI strategy false starts are the result of financial institutions failing to ask the right questions from the start:
- What real business problem are we solving?
- How will we test, iterate, and measure AI initiative performance?
- How does this improve customer experience, compliance, or revenue?
I've distilled these key questions and my conversations with leaders at the institutions getting the most from their AI initiatives today into the following best practices:
Anchor AI strategies to business goals
Tie every AI initiative to a clear business priority—whether it’s loan growth, customer retention, or operational efficiency.
Define KPIs, ROI targets, and adoption metrics before a project begins. No project should exist without a measurable path to value.
Start with the outcome you need, not the AI tool you want
The first step isn’t choosing a tool—it’s identifying the problem you need it to help solve.
When you begin with a business objective like expanding customer engagement, increasing loan volume, or improving retention, you can align your AI strategy with measurable outcomes. Otherwise, you’re investing in tech for tech’s sake.
Real example: When mortgage rates briefly dipped in September 2024, the Total Expert platform surfaced nearly $2 billion in refinance opportunities for our customers. But no human team can act on insights like that at scale. Our AI Sales Assistant enables personalized, timely outreach at scale—converting fleeting market shifts into new business.
Focus first on areas where a proof of concept or pilot is feasible within 30-60 days. Document classification, predictive churn modeling, or intelligent lead scoring are easy places to start, and those early wins build momentum, prove ROI, and prepare teams for more complex deployments.
Choose an AI solution that’s built for financial services
The reason so many financial institutions are frustrated is that they chose the wrong AI solution for their needs and industry.
Generic, horizontal AI platforms may work well in other industries, but they often fail in financial services—lacking the context, compliance, and data integration required. On the other hand, niche AI tools built for a specific function can’t scale to support financial teams.
Your AI solutions must be purpose-built for financial services and designed to scale at an enterprise level.
When vetting AI tools and solutions, ensure that they:
- Are compliance-ready (audit logs, consent, disclosure protocols)
- Have enterprise-grade security (end-to-end encryption, identity verification)
- Can scale quickly without compromising performance
- Will give you complete control over the outputs—you should never wonder what your AI might say to a customer.
To achieve scale and ROI, your AI solution must plug into your existing infrastructure—CRM, LOS, core banking systems, and customer data lakes.
Avoid building “AI silos” that require duplicating data or retooling your workflows. Instead, look for modular, API-driven architectures that extend what you already have.
Invest in data quality, governance, and compliance early
Your AI solutions will only be as good as the data you feed it.
Start by creating a single source of truth for customer and loan data. Then, anticipate obstacles to deploying AI with your data, such as consumer consent and preference management, and start addressing these things ASAP. Investing in tools like Customer Intelligence will help enrich your data and increase its value.
Regulations such the Gramm-Leach-Bliley Act (GLBA), TCPA (Telephone Consumer Protection Act), and UDAP (Unfair, Deceptive, or Abusive Acts or Practices) will be a few key areas where regulators dig in and look for companies cutting corners.
Think outside the chatbot box
Most people have interacted with a chatbot—and been frustrated by one.
That’s because traditional bots are scripted and rigid. They follow linear yes/no workflows. They can’t handle nuance, adapt to real-time questions, or navigate a human-level conversation.
Agentic AI is different. It learns from each interaction a develops the ability to shift topics, handle objections, gather missing information, and determine when the conversation needs to be handed back to a human on your team—will full context provided.
Unlike a chatbot, agentic AI knows each customer’s history, preferences, and financial situation, so it can have a two-way, brand-aligned conversation that feels human. That’s what creates trust, builds relationships, and drives revenue.
Train, educate, and empower your people
The most successful institutions are creating cross-functional AI task forces that give product, compliance, operations, customer experience, and data teams a seat at the proverbial table. They aren’t launching siloed pilots that only impact individual teams—they’re embedding AI into core workflows from the start.
This also means investing in change management and AI literacy—so that frontline teams understand, trust, and adopt the tools you’re giving them.
Choose the right tech partners: favor vertical specialists
Partner with vendors who understand financial services—especially your unique customer journeys or workflows. Deep domain understanding of core systems, database schemas, compliance, and other nuances will be a key factor in the results you achieve.
Benefits of vertical-focused partners:
- Deep understand of unique data sets and customer profiles
- Faster implementation with industry-specific models
- Built-in regulatory and risk controls
- Product roadmaps aligned to lending and banking trends
Horizontal AI tools have their place, but without deep domain expertise, they often require heavy internal customization and a slower time to value.
The future is here
I understand the urgency and pressure financial institutions are feeling to get their AI initiatives off the ground. Every day you delay, competitors are building up their capabilities, and you will struggle to catch up. As one of my investors put it bluntly, “Every day you fail to execute a comprehensive AI strategy, the value of your business goes down.”
But action without strategy leads to frustration.
To successfully launch and expand your AI initiatives, you need:
- A clearly defined business outcome for AI to support
- A purpose-built, enterprise-ready AI solution
- A cross-functional AI strategy that lays out your path to real value
When all that is in place, agentic AI won’t just optimize your business—it will open entirely new pathways for growth, scale, and customer connection. That’s how you move from buzzwords to business results.
To learn more about how Total Expert is working with our customers on high-impact AI initiatives, please reach out to our team.