Get a reality check on the AI in banking market

AI in banking is hot right now, and the back office is where it is moving fastest.

We've found a lot of interesting signals. For example, in February 2026, Goldman Sachs said it had spent six months building AI agents with Anthropic to handle trade accounting and client onboarding. And in May, Standard Chartered said it expects to cut more than 15% of its corporate-function roles by 2030 as it uses more automation and AI, which Reuters put at more than 7,000 jobs. The biggest banks are handing core back-office work to AI, and some already plan for fewer people.

In this section, we look at which banking jobs AI is taking on first, why the back office moved before lending decisions, and what these plans mean for vendors selling to smaller banks and credit unions.

Banks spend fortunes on work customers never see: checking new clients, reviewing alerts, reading loan files and answering the same questions every day.

We compare six kinds of AI banking vendors, from lending and credit tools to anti-money laundering, customer service, personalization and back-office automation. Their buyers range from global banks to community banks and credit unions. For each, we rank the pains that matter most, like money laundering alerts that are mostly false alarms, loan decisions that take days and call centers swamped by simple requests.

This shows which problems a bank will pay to fix now, and which ones it will wait for its core banking vendor to solve.

Licenses per bank or per seat, fees per loan decision or per alert, shares of the losses avoided, and AI sold inside a core banking contract: banking AI vendors get paid in several ways.

We compare who pays and how much it takes to serve each buyer. A global bank signs big contracts but takes a year to decide and often builds its own tools. A credit union pays less and needs the product to work out of the box. Fees per loan grow with lending, then shrink when rates rise. And every vendor pays for security reviews and model validation before the first dollar comes in.

You'll see which models grow as banks use them more, and which ones get absorbed when the core banking vendors add AI at no extra cost.

Banks have many reasons to spend more on AI, and some matter far more than others. We sort out which ones really move budgets.

The visible drivers are pressure to cut costs, growing compliance work and customers who expect instant answers. Others feed each other. Each loan decision a model makes adds data for the next one. When one large bank shows lower costs, its shareholders ask the others why they can't. And once a model risk team approves one AI tool, the second one tends to go through faster.

Before building or investing, you should know what convinces a bank to let AI touch money, customers or compliance, and what keeps it renewing after the first audit.

nCino and Upstart are the names many bankers know, but AI in banking reaches far beyond loan software.

A lot of the interesting work happens far from the bank's app. Some startups only review money laundering alerts for mid-size banks. Others read business tax returns for small business loans, answer the phones for credit unions, or predict which borrowers will fall behind before they miss a payment.

Comparing them is tricky. Banks rarely name their vendors, and many tools reach banks through core banking providers or consultants. Two startups with the same pitch can serve very different banks, one with a few branches and one with millions of customers.

Below, you'll find the market grids and ecosystem maps we've built to show which banking jobs each company handles, and where new players are breaking in.

Zest AI, interface.ai, Rogo… a few AI vendors have won real work inside banks and credit unions.

Zest AI helps lenders, many of them credit unions, approve more loans without taking more risk, and its own customers led its latest round. Interface.ai answers members' calls and chats for credit unions and community banks. Rogo, newer, does research, models and slide decks for investment bankers, and raised $160M at a $2B valuation in April 2026. Each found one clear job inside the bank, though none has yet shown it can cover a large bank end to end.

We'll look at how these teams chose one banking job to own, got through the reviews that stop most vendors, and widened their product from there.