The next several years in financial services will not be won by the company with the biggest roadmap.
They will be won by the company that can get elite AI builders inside the walls, give them authority, and make the operating model change.
That sounds obvious. It is not how most companies are acting.
Every legacy payments business I look at has some version of the same AI theater. Buy a tool. Run a pilot. Put a steering committee around it. Ask IT security to bless it. Wait six months. Produce a deck.
That is not transformation. That is procurement with a chatbot attached.
The real work is much harder and much more valuable: take the people who understand modern AI systems, put them directly against the workflows where money actually moves, and let them rebuild the company from the inside.
Invoice intake. Merchant onboarding. Underwriting. Exception handling. Risk review. CRM hygiene. Partner support. Reconciliation. Disputes. Compliance workflows.
These are not abstract productivity opportunities. These are hours, headcount, error rates, customer response times, and EBITDA.
I have been living this in my own business. Fern Capital is a one-person company, but the operating team is increasingly a set of AI agents. Last week, an automated review found six real problems across my systems and wrote them into a tidy document. Nothing moved. So I changed the rule: problems now land in one visible queue, and three nights a week an AI agent picks the oldest one and fixes it. A second AI, from a different vendor, reviews the change, and an unresolved objection blocks the merge.
The first night I armed it, the reviewer caught five design flaws in the fixer itself.
That is the point. The leverage did not come from buying AI software. It came from redesigning the work, deciding where judgment lives, and giving the builder enough authority to change the machine.
That is why talent is the whole thesis.
M&A should be viewed through that lens. A small AI-native product team with real payment workflow knowledge may be more strategically important than a larger company with more revenue but no technical leverage. The question is not just what revenue comes with the asset. The question is what capability comes with it.
Can this team actually build?
Can they integrate into legacy systems?
Can they automate work that still lives in inboxes, spreadsheets, PDFs, and swivel-chair operations?
Can they get adoption from operators who have spent twenty years doing the job one way?
That is where diligence has to go.
The diligence checklist is not complicated, but it is different from the usual deck review. Show me a production workflow the team changed. Show me who has access to the data and the systems. Show me how work gets tested before it touches customers or money movement. Show me whether operators use it because it helps them, or because management told them to.
Most buyers still underwrite the software. I would underwrite the builder.
The uncomfortable part is that a lot of people inside these companies will not make the transition. I have seen this pattern enough times to say it plainly. Some people can be retrained. Some can be moved. Some protect the old process because the old process is where their authority lives.
That does not mean you rip out institutional knowledge and hope the model figures it out. That is how operations break. The right move is staged: pair the builder with the operator, document the workflow, move the first low-risk queue, measure the result, then expand authority as the evidence builds.
AI is not a feature layer. It is a labor model change.
And if the leadership team is too far from the work to understand that, the company will move too slowly. Low-level employees will get blocked by old IT processes, security reviews, internal politics, and managers who do not understand what is possible. The best builders will leave. The pilots will die. The board will get a quarterly update that says progress is being made.
Meanwhile, someone else will be rebuilding the actual operating system.
That is the opportunity in payments right now. Not AI as a press release. AI as an operating advantage.
The winners will have three things: builders with real authority, executives close enough to the work to know what should change, and boards willing to back the disruption before the P&L makes it obvious.
Everything else is noise.
Question for payments operators and investors: when you diligence an AI story, are you underwriting the software, the workflow change, or the talent that can actually deliver it?
And for anyone running one of these businesses: do you know where your best AI builder is spending their time this week?