Boston Fintech Week 2026: What Credit Unions Can Learn from Banking's Debate Over Agentic AI
By Tylor Tourville
Agentic AI was one of the most prominent themes at Boston Fintech Week 2026, with discussions exploring its potential to transform financial services alongside growing questions about trust, accountability and the risks of allowing AI systems to autonomously act.
One session, titled “The AI Agent & the Bank: Credit, Cores, and Compliance,” tackled those questions head-on. The discussion featured Peter Hazlehurst, Co-Founder and CEO of Synctera, and Ashley Nagle Eknaian, Chief Digital Officer of Eastern Bank, and moderator Kimberly Monty Holzel of Goodwin Procter.
Hazlehurst was quick to acknowledge the stakes involved, expressing skepticism about how much autonomy financial institutions should entrust to AI agents. “The consequences of getting this wrong is very high,” he said.
For credit unions, these questions are becoming increasingly relevant. According to Cornerstone Advisors' What's Going On in Banking 2026 survey, 17% of credit unions have already invested in or deployed agentic AI, while another 25% planned to invest or implement the technology this year.
Financial Institutions and Fintechs Are Approaching AI From Different Directions
The panel highlighted a growing disconnect between fintech companies rapidly integrating AI into their operations and financial institutions operating within more demanding risk and compliance frameworks.
Hazlehurst, whose company Synctera helps connect fintechs with regulated financial institutions, sees this tension firsthand. “When it gets complicated is when the bank says they can't use that, but the fintech says, we only use that,” he said. “Banks generally have a higher bar when they report what's happening to your data. Fintechs don't understand that part or haven't had to think about it.”
However, Hazlehurst suggested that both sides have something to learn from one another. While financial institutions bring deep expertise in risk management and regulatory compliance, fintechs can help challenge assumptions about what constitutes an acceptable level of risk. “The domain expertise will always be there, but the fintechs are helping the banks understand that this isn't as risky as they think it might be,” he said.
For credit unions evaluating AI solutions from fintech partners, the discussion underscores the importance of balancing innovation with appropriate oversight, rather than allowing either objective to dictate the relationship.
Who Is Responsible When an AI Agent Makes a Mistake?
Another central issue raised during the panel was the distinction between traditional automation and autonomous AI agents. Historically, financial institutions have relied on deterministic systems that follow predefined instructions. Agentic AI introduces a different level of complexity, with systems capable of independently interpreting information and determining how to act.
Hazlehurst offered an example illustrating the potential consequences. “Let’s say Anthropic scrapes my bank's website. How would I know if they got something wrong? And if it does go wrong, who is accountable?”
The discussion also raised questions about authorization and dispute resolution when AI agents initiate financial transactions on behalf of consumers. “Going forward, we just don't have a model for tracking who said what to whom,” Hazlehurst said.
For credit unions, these scenarios raise important questions about how existing fraud prevention, identity verification and dispute resolution processes will accommodate AI agents acting on behalf of members.
AI Adoption Will Require Stronger Guardrails, Not Just Better Technology
The conversation also turned to the safeguards financial institutions will need as AI becomes more autonomous. Eonavian emphasized the importance of involving stakeholders beyond technology teams when developing AI applications, including legal, compliance and risk professionals. “You don't want just engineers working on something. You have to have experts around the table,” she said.
Hazlehurst also raised concerns about whether defensive technologies are keeping pace with the capabilities of AI systems and the threats they introduce. “A whole category of technology is yet to be built, which is defensive,” he said, adding that the defensive side is not nearly as advanced as the offensive side.
For credit unions, the implications extend beyond evaluating the capabilities of an AI solution. Responsible adoption will also require assessing the governance, security controls and operational maturity of the technology partners providing it.
Consumers Care About Outcomes, Not the AI Behind Them
Despite the risks and challenges, the discussion ultimately returned to what people expect from their financial institutions.
“I don't think consumers think, what can AI do for me? They're thinking of a problem they want to solve,” Hazelhurst said.
For credit unions, the challenge will be delivering on AI's promise of a more convenient member experience while preserving the trust that underpins their relationships with members.
