Last week, OpenAI quietly integrated ChatGPT with Plaid. This is something that should get the attention of every community bank and credit union executive in the country.
Customers, members, and businesses can now connect their financial accounts directly inside ChatGPT, and ask the AI to analyze spending, track subscriptions, model financial scenarios, and assess investment risk. The tool launched for ChatGPT Pro users and supports more than 12,000 financial institutions in the Plaid network. An Intuit integration is already planned. The feature is expected to roll out to Plus users in the near future.
This is not a side project. It is a deliberate, well-resourced move into personal financial management territory that banks and credit unions have been building toward for years.
And it is only one side of a squeeze.
The Top of the Squeeze: Consumer AI That Doesn't Need You
For the past several years, the dominant question in banking technology has been: how do we give customers a better digital experience? The assumption behind that question is that the bank owns the digital relationship. The customer comes to the bank's app, the bank's website, the bank's branded interface.
OpenAI just challenged that assumption directly.
When a customer connects their accounts to ChatGPT and starts asking financial questions, they are not going to the bank's interface. They are going to a general-purpose AI that happens to have read access to the bank's data. The bank becomes a data source. The AI becomes the relationship.
This matters because the value of being a community financial institution has always been built on proximity and trust. You know your customers. You understand their financial lives. You are local, you are accessible, and your engagement is not commoditized.
A well-built AI that can see someone's full financial picture, answer questions intelligently, model scenarios, and remember preferences does not care that you have a branch on Main Street. It is available at 2 a.m., it does not put you on hold, and it synthesizes information across every institution a customer uses, not just yours.
The expectation gap this creates is real. Customers who use these tools will start to compare their experience with their bank to their experience with ChatGPT. That comparison may not always favor the bank.
The Bottom of the Squeeze: Core Providers Embedding AI into the Infrastructure
While consumer AI platforms are building toward the customer relationship from the top, the core banking providers are building toward it from the bottom.
FIS, Fiserv, and Jack Henry, the three companies that collectively power the back end of most community banks and credit unions in the country, are all integrating AI into their core platforms. FIS is applying AI to fraud detection and personalization, leveraging 200 petabytes of transaction data from their installed base. Jack Henry is on track for an H1 2026 launch of a public cloud-native core and continues to modernize its banking infrastructure. Third-party platforms like AgentFlow are building integrations that connect agentic AI directly to these core systems for document processing, compliance, and decisioning workflows.
The strategic implication here is different from the consumer AI threat, but equally significant. When AI capabilities are bundled into the core, the institution's ability to differentiate on technology is constrained by what the vendor chooses to build and sell. You get AI, but you get the same AI as the 500 other FIs on the same core.
Differentiation through technology, which used to require custom development, increasingly requires either switching cores or layering specialized tools on top of a commodity infrastructure.
The Middle: The Fintech Vendors Selling Bank-Branded AI
Companies like Glia, or interface.ai with their Sphere and BankGPT platforms, are selling AI directly to financial institutions for customer-facing deployment. Their pitch is that community banks and credit unions should have their own AI, one that carries the institution's brand, is trained on their products, and lives inside their customer engagement channels rather than inside a general-purpose chat tool.
Interface.ai reports close to 100 financial institution clients, 500 million total conversations processed, and recently closed a $30 million funding round that positioned them as the most well-capitalized agentic AI company specifically focused on banking. Their Smart Collections product, launched in 2026, targets delinquency management across phone, SMS, and chat channels. Eltropy is competing in the same space.
This is a legitimate response to the consumer AI threat. An FI-deployed AI agent that a customer trusts and interacts with regularly does establish and reinforce an institutional relationship. The problem is that it requires meaningful investment, vendor selection discipline, and integration work that many community institutions are not yet equipped to execute quickly.
And it does not solve the core provider commoditization problem. You can deploy interface.ai on top of Jack Henry, but if the AI landscape consolidates and the core providers build comparable capabilities into their platforms natively, the differentiation window closes.
The Real Question: Who Owns the Financial Relationship in Five Years?
OpenAI is betting that the AI layer will become the primary interface through which people manage their financial lives, with banks reduced to licensed data providers and transaction processors. They may be wrong, but the bet is credible enough that it is being funded at scale.
Core providers are betting that deep integration with back-end systems gives them a defensible position to deliver AI that actually executes, not just advises. That bet is also credible.
Fintech vendors in the middle are betting that institutions will pay a premium for AI that is institution-branded and institution-controlled. That bet depends on whether "branded" matters enough to customers, and whether smaller institutions have the budget and technical capacity to adopt and operate these systems well.
For community banks and credit unions, the honest answer is: the outcome is not determined yet. But the time available to build a coherent strategy is shorter than it was a year ago.
What Institutions Should Actually Do
Understand your data posture. Any AI strategy, whether you build, buy, or partner, depends on clean, accessible, permissioned data. Most community institutions do not have this in order. That is the foundational work, and it is boring, but it is what separates institutions that can adopt AI from institutions that can only watch vendors demo it.
Do not wait for your core provider to solve it. Core modernization is slow. The AI features your core vendor ships in two years will reflect decisions they are making today, and those decisions are not being made with your specific customer base in mind. Institutions that are building a strategic view now will have better vendor leverage and faster adoption when capabilities arrive.
Evaluate the engagement threat honestly. Pull your customer interaction data. Where are customers spending time in your digital channels, and what are they not doing there that they should be? If your mobile app engagement is thin, that is where a consumer AI with full account visibility starts to substitute for you. Fixing engagement before the substitution accelerates is a better position than fixing it after.
Be selective about fintech vendor commitments. The middle layer of bank-branded AI vendors is active and growing. Some of them will be around in five years; some will not. Institutions entering multi-year contracts in this space should be doing reference checks, asking hard questions about data portability, and understanding what happens if the vendor is acquired or pivots.
The move OpenAI made with Plaid is not a crisis. Community banks have survived many rounds of technology disruption, and the local trust and relationship advantages that define the best institutions are real.
But the structure of the competitive environment is changing, and it is changing from two directions simultaneously. The institutions that understand that structure clearly, and build strategy around it rather than around any single vendor's pitch, are the ones that will be in a strong position when the landscape settles.