As people turn to AI for money guidance, institutions can keep those conversations grounded in trusted relationships. Josh DeTar discusses building useful AI with accuracy, transparency and human judgment.
Trust has always been fundamental to banking. Why does it feel especially consequential for banks and credit unions today?
TD Bank found 55% of Americans had used AI to manage finances, up from 10% a year earlier, though only 18% trust AI to recommend on its own. OpenAI says more than 200 million people ask ChatGPT money questions monthly. Institutions already hold the data and trust people want. Ceding that conversation to AI Assistants can cost them business and erode public perception over time.
Credit unions have long differentiated themselves through relationships and community. How can institutions preserve that advantage as more member interactions move into digital channels?
AI could leave a credit union holding the deposit while another service holds the relationship. Local trust matters: 35% cite community involvement as a reason to switch institutions, RivalMind found. But even though locality matters, convenience often wins. But if you pair community connection with AI-enabled tools that make everyday tasks easy, you've got a winning combination.
As AI becomes part of how consumers seek advice, make decisions and manage their money, how might it change the relationship between a financial institution and the people it serves?
Banking is shifting from displaying balances to offering guidance. Asking whether they can afford a home or are overspending on a car can deepen the relationship. Asked elsewhere, the institution becomes someone else’s data source. Deloitte found 79% of people trust information on their bank’s website, compared with 46% who trust banking recommendations from generative AI.
Many institutions are moving quickly to introduce AI capabilities. What should they consider about transparency, reliability and consumer confidence before bringing those tools into the banking experience?
Ask whether AI provides faster, more relevant information or helps complete a task. Ground answers in approved content and account data, and don’t be afraid to admit uncertainty. Disclose AI use, keep a person available and always require approval before action.
Where do you see the greatest risks to trust as AI adoption accelerates - whether in data privacy, security, accuracy or the loss of human connection?
For me, inaccurate answers are still the greatest risk. Data exposure is another; recently a tech company CEO’s AI agent posted his bank balances and spending to the company’s general Slack channel. This is why the trust element is crucial: a governed institutional environment with regulatory and compliance standards can dramatically reduce that risk.
Could AI actually make banking more human? Where can automation create more space for meaningful member interactions, and where is human judgment still essential?
Absolutely. AI can handle preparation and note-taking, making room for human attention and less judgmental money questions. At one credit union, a member asking how to save for a first home prompted staff to follow up; that member is now buying a home. AI can surface needs, but people should be the ones to make consequential decisions, especially about lending, hardship and major life events.
Tyfone is investing in AI-enabled digital banking, including nFinia and Fathom. What problems are you focused on solving, and how does trust shape your approach to building these capabilities?
Fathom, our AI, is built into nFinia, not bolted on as an FAQ. Members can ask how to deposit a check, analyze finances or create a widget tracking a spending category. Answers use approved institutional content and member data; sensitive information is masked, data is not used to train outside models, interactions are auditable and no money moves without member approval.
Personalization can make banking more helpful, but it can also feel intrusive. How should financial institutions strike that balance?
Ask rather than guess. Predicting needs from transactions can feel intrusive. If a member says they hope to buy a home and asks how to prepare, the institution can offer relevant help with permission. That responds to a stated goal, not an assumption.
For a bank or credit union leader planning the next phase of AI adoption, what practical steps would you prioritize to ensure innovation strengthens trust rather than erodes it?
Test information, efficiency and trust: does the tool improve accurate information, make a task easier and pass what the institution can stand behind? Bring compliance in before choosing vendors. Start with navigation and spending insights and require confirmation for actions. Seek independent audits, assurances that customer data will not train models and audit logs. Keep people responsible for consequential decisions.
Fintech Meetup 2027 is centered on the theme 'Built on Trust.' What conversations should banks, credit unions and fintech partners be having now to build a more trustworthy financial ecosystem?
Ask whether partners will preserve account-holder relationships, meet compliance needs and deliver real value. The industry must agree how AI agents are authenticated, what they can do and who is responsible when they fail. Plaid found 80% of consumers expect companies to reimburse AI-driven mistakes. With the CFPB’s open banking rule still being reworked, set accountability standards now. Build AI inside trusted relationships, not around them.