Ready to make confident, compliant, and strategic AI decisions for your credit union? Watch the recording of the "Responsible AI: Vet Vendors, Build Consensus" webinar to get the playbook you need. Keep reading for a summary of the conversation and a checklist to use in your search.
There's a lot of talk about AI today, and if you're a leader in a financial institution, you're likely feeling the pressure to figure out your strategy. The good news is, you're not alone. The vast majority of banks and credit unions are on this same journey, trying to make sense of the overwhelming number of AI solutions and pitches.
So, how do you separate the hype from the solutions that will actually deliver value, securely, and in a way that aligns with your institution's core values?
That was the central question of our recent webinar, "Responsible AI: Vet Vendors, Build Consensus." We brought together Elizabeth Thornton, VP of Project Development at Gulf Coast Educators Federal Credit Union, and Justin DiPietro, Glia's co-founder and Chief Strategy Officer. They had an open and honest conversation about adopting AI, providing a practical rubric for evaluating AI vendors. This guide will help you make the smartest, most responsible choices for your institution.
Here are the questions you need to ask when evaluating AI vendors.
1. Does the AI Vendor Understand Your Industry? 🎯
This is the most critical starting point. A generic AI solution designed for a pet food provider can't possibly meet the stringent security, compliance, and nuanced needs of a bank or credit union. Glia's Justin DiPietro emphasized this point by explaining that AI for a regulated industry must be built with a completely different mindset.
This distinction is vital because general-purpose AI systems designed for broader applications might permit undesirable behaviors, such as hallucinations or less stringent data privacy practices—both unacceptable in the highly regulated financial sector. DiPietro further explained that banking and credit union interactions often involve unique complexities, such as multi-level authentication, where the required verification varies significantly based on the risk level of the transaction, from a simple balance check to a large fund transfer.
This specialized understanding allows for the development of AI solutions that can deliver specific value, like automating low-risk inquiries, while maintaining the necessary security and compliance, ensuring that customers aren't subjected to unnecessary authentication steps for simple requests.
Elizabeth shared that Gulf Coast Educators FCU, like many financial institutions, is excited about being "digital first," but they're also "moderately apprehensive" about the security of their members' data. They needed a partner who understands that concern from the ground up, not one who retrofits a general-purpose tool.
❓To determine if a vendor is purpose-built for your industry, ask these questions:
- Can I speak with another bank or credit union using your AI tools?
- How do your AI tools support multi-level authentication requirements?
- Can your AI tool ensure verification steps differ based on the level of transaction risk?
- How do you monitor changing regulations from agencies such as NCUA?
✨Our Advice: AI should be ‘Banking-First, not Banking-Ready’. You need AI that is tailored to your use cases, from fraud prevention to automated balance inquiries, ensuring immediate value without the added risk or complexity of a generic platform.
2. Is the AI Truly Human-Centric? 🤝
For many, the idea of introducing AI can feel like a direct threat to the human touch that is so critical to building trust and loyalty (and makes community banks and credit unions so special).
Historically, institutions have felt forced to choose between efficiency and experience. On one hand, you can optimize for cost savings with automation and fewer staff, but this leads to poor experiences for customers. On the other hand, a focus on high-touch, personalized service can lead to skyrocketing operational costs. This perceived "trade-off" is at the heart of the hesitation around AI adoption.
Elizabeth emphasized this key concern for her team: the fear that AI might compromise the human touch that defines their brand. But from the start, her team had a bigger goal with AI—one that would ensure the human touch remained prominent in their member interactions.
“Our biggest goal is removing barriers and manual tasks to make sure we’re maximizing our team’s resources and the personal touch with members. So, if we can automate processes or alleviate friction points with AI, then we can refocus those resources on very important touchpoints.”
Justin echoed this sentiment, highlighting that a significant portion of calls to banks and credit unions often consist of the same simple queries like "what's my balance?" or “I need to order new checks.” When AI can automate the routine, agents can be reallocated to elevate other areas of the business. One example he shared was a bank that refocused its staff from answering repetitive queries to engaging in “revenue-generating activities” like CD renewals or HELOC offers.
At GCEFCU, Elizabeth’s team has been able to focus on connecting with members since implementing Glia AI. She shared: “That’s one of our main focuses, and continuing to reboard members that have been members for a while and may not be familiar with our latest products and services.”
The most valuable AI solutions are those designed not to replace human agents, but to empower and elevate their roles, ensuring that human expertise is reserved for complex, empathetic, or revenue-driving conversations.
❓To determine if a vendor’s AI will support or mitigate the human elements of your business, ask:
- How do you determine the best use-case for each type of AI used in your tools?
- Do you measure the success of AI with Containment Rate (How well an interaction is contained by AI) or Understanding Rate (How well the AI understands the inquiry)? If the AI truly understands the query, the organization retains control over whether to contain the interaction or escalate it to a human, rather than forcing containment for the sake of a metric.
✨Our Advice: You need a tool that allows every contact center stakeholder to unlock the value of AI at scale. The goal is to automate routine tasks and free agents to focus entirely on complex, empathetic interactions—the very moments that build strong customer relationships and drive loyalty. But this idea shouldn’t stop at agents. Ensure your AI partner has features available that elevate every function of the contact center, like reporting, workforce management, and real-time coaching.
3. How Robust is Security and Data Governance? 🔐
For financial institutions, security is non-negotiable. But asking a vendor if they are "secure" is like asking a used-car salesman if the car runs well. You need to know the right questions to ask.
Justin recommends looking beyond common certifications like SOC 2 and ISO 27001, which are often "checkbox security". Instead, he suggests a vendor's PCI Level 1 certification as a better indicator of true data security, as it requires a much deeper audit of their technology and operations. Another important question is where your data is physically going. Is it leaving the vendor's infrastructure to train a third-party model? This is a critical point of exposure.
Justin highlighted that even if a vendor claims they don't use customer data for training, or if a checkbox is selected on a site to prevent it, there's often no guaranteed certainty unless the vendor deploys and controls the models within their own infrastructure.
At Glia, we train and deploy AI models behind our own infrastructure so that we know we're 100% certain that they’re never trained using customer data, whether they are proprietary or open source models.
This level of control is crucial for financial institutions. Without knowing where your customer data is going, the data can end up in the wrong place, posing significant risks to data privacy and security.
Our Advice: AI models should be hosted and deployed within their own secure infrastructure. This means your data is never used to train third-party models, guaranteeing privacy and compliance. Make sure your AI platform also offers robust security features like PII masking and data redaction, ensuring the utmost security for your digital interactions.
❓To determine if a vendor will uphold your robust security and compliance requirements, ask:
- What models are you using? Where are they hosted? How are they trained?
- How are you sure that our member/customer data is not being used to train third-party models?
- Where does our member/customer interaction data go? Does it go through a third party?
- What source material are you using to train AI voices? This can help avoid potential copyright or regulatory issues.
- Do you have industry relevant certifications, such as SOC-2 and ISO 27001?
- Do you have PCI certification (specifically PCI Level 1)? Many vendors rely on their cloud provider’s certification, but this difficult to obtain certification is a strong indicator of the robust security banks and credit unions need.
- Do you have a bug bounty program? Such a program signals a company's confidence in its security, as it invites external ethical hackers to find vulnerabilities. A lack of such a program might indicate an unwillingness to expose potential security weaknesses due to legal risks.
4. Is the Vendor a True Partner or Just a Salesperson? 🤝
Elizabeth’s journey at Gulf Coast Educators FCU led her to a key piece of advice: "Hold your vendors accountable". Instead of just listening to a pitch, you should share your strategic goals and ask them how they will help you achieve them. It's also important to check a vendor's track record. AI is moving fast and vendors need to not only be reliable in their commitments but also agile and forward-thinking enough to continually integrate the latest technological advancements.
“It’s really easy to build a product roadmap,” Justin explained, “but it’s a lot harder for vendors who say they’re going to do a lot to go back and show you what they actually delivered last year.”
❓To truly assess a vendor's ability to execute and adapt, he recommends a more discerning approach:
Request Past Roadmaps and Actual Deliveries: Justin advises, "When I evaluate vendors myself, I always say, ‘I want to see a PDF product roadmap from one year ago.’ And then I want to see what they've built since then, because I want to know how accurate the roadmaps are."
This practice helps verify if the vendor consistently delivers on their stated plans.
Verify Production Readiness: He also cautions against being swayed by impressive demos that are not yet in production. He noted, "I cannot tell you how many times I see demos and it's not actual production, or sometimes it's just designs". It's crucial to confirm that demonstrated features are truly live and functional.
✨Our Advice: When it comes to AI, you need an uncommonly great partner, not just a vendor. Ask for existing customer references in your industry, and don’t forget to look at customer reviews on G2.
The Future is Now 🔮
The AI era is here, and the potential for a seismic impact on the front-line labor force is undeniable. Glia’s co-founder, Justin DiPietro, summed it up best by saying, "The only reason you do these things, the only reason you even ask security or compliance questions, it all comes down to efficiency and value. When your organization adopts AI, you want to be sure you are creating a safe and responsible foundation for long-term innovation."
Choosing the right AI partner means more than just adopting new technology; it's about making a strategic decision that affects your team's morale, your members' experience, and your institution's long-term growth.
Elizabeth's proactive approach at Gulf Coast Educators Federal Credit Union is a testament to the leadership required in the AI era. By developing a clear rubric and demanding transparency from vendors, institutions can move past the hype and make confident, data-driven decisions. The right AI partner provides more than just a product; they offer a roadmap for success, giving you the strategic agility to lead your organization forward and a clear path to turn your contact center into a key driver of value and growth.
Ready to make confident, compliant, and strategic AI decisions for your credit union? Watch the recording of "Responsible AI: Vet Vendors, Build Consensus" to get the playbook you need.
Ready to see how Glia's AI for All™ can help you automate routine tasks and elevate your team to deliver exceptional, cost-effective service. Request a personalized demo today!
.png)
.avif)
.avif)
