WHO KNEW? Issue 1 | May 2026 | Artificial Intelligence in Payments — Agentic AI
Welcome to WHO KNEW?, a new monthly series from NEACH that examines a single, significant development in payments — the kind of shift that warrants a closer look. Each issue follows a consistent structure: one key development, context to help make sense of it, and practical next steps for your institution. This month’s topic: Agentic AI in Payments and what financial institutions are already doing with it.
THE “WHO KNEW?” MOMENT
Agentic AI in banking is becoming the new norm.
Consider that BNY — the world’s largest custody bank — reports that it now has more than 20,000 employees actively building AI agents on its internal platform, Eliza, with over 125 use cases in production. The bank has created more than 100 “digital employees” — AI agents with their own bank user IDs, email accounts, and system access — to handle tasks such as payment validation and code security. One agent reduced legal contract review time by 75%, from four hours to one, across 3,000 vendor agreements each year.
Globally, BNY is not alone. Crédit Agricole Bank Polska indicates it used agentic AI to cut document-processing time in half, saving more than 750 hours per month2. Brazil’s Bradesco notes that it freed up 17% of employee capacity and reduced lead times by 22% by prioritizing agentic use cases in fraud prevention and customer service3.
Q1: What exactly is Agentic AI, and how does it differ from other forms of AI?
Most financial institutions are already using two types of AI. Generative AI (such as ChatGPT or Microsoft Copilot) creates content — drafts, summaries, and analysis — based on prompts. Predictive AI analyzes historical data to identify patterns and forecast outcomes, such as flagging suspicious transactions or scoring credit risk.
Agentic AI operates differently. As described by Nacha’s Payments Innovation Alliance in Artificial Intelligence: Rapidly Redefining the Payments Landscape, agentic AI is proactive — it pursues a goal with minimal human involvement, determines its own plan, uses the tools it needs, and adapts as conditions change.
In practical terms, generative AI is a tool practitioners use. Agentic AI functions more like a worker that can be directed and managed.
Q2: What does this mean for community and mid-sized institutions?
The examples above involve large institutions, but agentic AI’s reach extends beyond them. As ICBA’s Independent Banker notes, most community banks will not build their own AI platforms4. Instead, they will access this capability through products already purchased or leased from core providers and vendors, which is the most realistic on-ramp for smaller institutions.
The use cases with the most immediate relevance include:
• ACH return items, where agents can automate resolution decisions.
• Fraud prevention, where agentic AI can flag and stop transactions in real time rather than after the fact.
• Small-dollar credit underwriting, in which traditional, human-intensive review is applied to every consumer loan and is cost-prohibitive. AI agents meeting defined criteria could handle less complicated approvals.
• BSA/AML compliance, where early-adopting banks are reporting significant reductions in onboarding time and staff workload on Know Your Customer (KYC) tasks.
Q3: Where should institutions begin?
A reasonable first step is to understand what core providers and fintech partners are currently doing with this technology — because the tools are likely arriving through channels institutions already use.
NEACH’s monthly Executive Summary tracks these developments and highlights how financial institutions are using AI.
WANT TO GO DEEPER?
NEACH’s 2026 Future of Payments Symposium is scheduled for November and will bring together financial institution leaders, fintech builders, and payments strategists to examine AI adoption, emerging payment technologies, and strategies for building teams capable of acting on innovation.
Watch fps.neach.org for speaker and session announcements.
---
[1]Source: https://openai.com/index/bny/. Accessed 4/22/26.
2Source: https://deviniti.com/software-development-case-studies/case-study-ai-agent-at-credit-agricole/. Accessed 4/22/26.
3Source: https://www.mckinsey.com/industries/financial-services/our-insights/banking-matters/digital-banking-speed-scale-and-the-agentic-arms-race. Accessed 4/22/26.
4Souce: https://www.independentbanker.org/w/understanding-the-opportunities-and-concerns-for-agentic-ai Accessed 4/28/26.