Well, it's official. Wall Street banks are rolling out AI assistants. Not for fun, because that's not what Wall Street does. They're doing it for productivity and more efficiency. They're going to use it for client vetting, treasury, trading, wealth management, and internal operations. And again, there's another signal. AI is moving from chat to actual work in real scenarios. Every industry is about to ask the same question. Where are we still making people manually move information from one place to another? That's where the opportunity is. That's what I keep saying. No more software, less friction, more efficiency.
Let me sit on this, because the specifics of what these banks are actually doing are more interesting than the headline, and they tell you exactly where this is all going.
Reuters reported that major banks are now racing each other on agentic AI, meaning AI that completes tasks with minimal human supervision, across wealth management, client vetting, trading, and treasury. Goldman Sachs partnered with Anthropic to build agents that handle trading, transaction accounting, and client onboarding. JPMorgan is targeting corporate treasury. Morgan Stanley starts testing client-facing assistants this summer. UBS has agents firing thousands of daily alerts to advisors and executing trades and transfers once the advisor signs off. A KPMG survey found 51 percent of banks are already piloting AI agents.
And then there's BNY, which is the detail I can't stop thinking about. They give their digital employees login IDs. Nicknames. A human manager responsible for training them. Quality control. Something like a performance review. One of them is named Payment Pete.
That's not a chatbot. That's a coworker with a job description.
Now, understand who's doing this. Wall Street is not a place that adopts technology for the fun of it, or to look modern, or to have something to say on an earnings call. Banks are among the most risk-averse, most regulated, most conservative institutions in the economy. When they move, they move because someone did the math and the math was undeniable. Investors are asking exactly one question about AI spend right now, which is where the return is, and the banks are pointing at these functions because that's where the return is legible.
So this is a signal, and it's a specific one. AI is moving from chat to actual work in real scenarios. The demo phase is over. The pilot phase is ending. We're now in the phase where an AI has a login and a task list.
But here's the part I want every person in every industry to actually take away, because it's not "Wall Street is using AI." That's the headline, and headlines are cheap. The real question is why those specific functions. Client vetting. Treasury. Trading. Onboarding. What do they have in common?
Every one of them is a place where a human being was manually moving information from one place to another.
Client vetting is taking information from a client and moving it into a compliance system. Onboarding is taking information from a form and moving it into an account. Treasury is moving numbers between systems that don't talk. Transaction accounting is taking what happened and recording it somewhere else. These are not intellectual tasks. These are transport tasks. A human reads something here, and types it there, and the work has zero judgment in it, and it eats the day.
That's the pattern. That's the whole pattern. Wall Street didn't ask "where can we use AI." They asked "where are we paying skilled people to move information by hand," and they pointed the agents at those places.
So every industry is about to ask the same question, and most of them haven't started yet. Where are we still making people manually move information from one place to another? Where is a professional we pay for judgment spending their afternoon retyping something that already exists somewhere else?
Once you start looking for it, you find it everywhere. It's in the intake form that gets re-entered into three systems. It's in the notes from a conversation that get retyped into a record. It's in the document that gets rebuilt because the information lives in a format the next step can't read. Every industry has these, and every industry has been treating them as the cost of doing business, and Wall Street has decided that they aren't.
That's where the opportunity is. Not in adding another piece of software. That's what I keep saying and I'll keep saying it. No more software. Less friction. More efficiency. The winners in every field are going to be the ones who look honestly at their own day, find the places where humans are acting as transport for information, and hand those to a machine so the humans can go back to the parts that actually need them.
Wall Street just showed everybody where to look. The rest of us should be taking notes.
*Judd Hoffman is CEO and Co-Founder of Ethica AI, building AI-powered tools for real estate transaction workflows.*