Innovation is a performance for an audience that doesn’t actually do the work. We like to think that corporate priorities are determined by a cold, mathematical assessment of Return on Investment, but that assumes the people making the decisions are looking at the same balance sheet as the people on the floor.

Most of the time, they aren’t. They are looking at a stage, and they are worried about whether the lighting makes them look modern or like a relic of the late nineties.

I spent forty minutes this morning staring through my car window at my keys, which were resting on the driver’s seat. It was a stupid, human error-the kind of friction that shouldn’t exist in a world of proximity sensors and smart locks.

But it happened because I was thinking about three different things at once, none of which involved the physical reality of a piece of metal and a deadbolt. Corporations do this every day. They lock themselves out of their own efficiency because they are too busy thinking about the “strategic narrative” to remember where they left the keys to the actual operation.

The Digital Disconnect

This disconnect is never more visible than in the current rush toward Enterprise AI. We are witnessing a bizarre era where a servicing manager in a mid-sized equipment finance firm is pulled into a six-hour “Innovation Workshop” to brainstorm generative AI use cases, while three of her best employees are currently sitting in a windowless room manually reconciling lockbox files because the core system doesn’t talk to the bank’s reporting portal.

The workshop is filled with sticky notes-neon pink and lime green-representing a future where “intelligent agents” handle customer inquiries. The facilitator, a person whose clothes have never touched a day of actual portfolio administration, asks the room to “dream without constraints.”

The servicing manager writes down a few buzzwords. She has to. It’s part of her KPI now. But as she walks back to her desk, she passes Dana.

Dana has two monitors. On the left is a PDF of a remittance advice from a construction company that just leased twelve backhoes. On the right is a spreadsheet that looks like a digital representation of a migraine.

Dana is manually matching a $42,840.12 payment to seventeen different contract line items because the customer’s accounting department decided to roll three late fees into a single lump sum without telling anyone.

The enterprise has allocated a $2 million budget to the AI initiative. The request to automate the remittance matching was denied six months ago because it was “not a strategic priority at this time.”

This is the central paradox of modern finance: The gap between the announced agenda and the actual work is a reliable measure of who the agenda is for.

If the work is for the customers and the employees, you fix the remittance matching. If the work is for the board of directors and the press release, you build a chatbot.

The Oliver Evans Problem

In the late 18th century, a man named Oliver Evans built what was arguably the world’s first fully automated factory. It was a flour mill in Delaware. Through a series of buckets, conveyors, and Archimedean screws, grain entered the mill and emerged as fine flour without a single human hand touching it.

It was a marvel. It was also a commercial disaster at first. Why? Because the millers who saw it didn’t trust it. It didn’t look like work. To them, “work” was a man covered in white dust, sweating over a grinding stone. If there was no visible struggle, they assumed the quality was poor.

We have inverted Evans’ problem. Today, we want the appearance of automation without doing the hard, grinding work of fixing the data architecture. We want the “AI” label because it signals to the market that we are “future-ready,” even if the “future” is being held together by Dana and her spreadsheets.

This isn’t just a matter of poor prioritization; it’s a failure to understand the difference between visible and invisible labor. AI is visible. It’s a talking point. It’s something a CEO can mention in an earnings call.

Manual remittance matching is invisible. It’s the “plumbing” of the equipment finance world. Nobody gets a promotion for making the plumbing work 10% faster, but people get fired when the pipes burst and the delinquency roll rates start climbing because payments weren’t posted on time.

The servicing manager knows this. She lives in the tension between these two worlds. She knows that if she could just get her data into a governed, API-first environment, she wouldn’t need a “dreaming workshop.” She would just need a system that functions.

When we talk about equipment leasing software in the modern context, we are often talking about bridge-building. In the United States, the equipment finance market is uniquely fragmented.

You have bank-owned lessors with legacy core systems that feel like they were programmed on a loom, and you have independent captives trying to move at the speed of a tech startup. Both are currently being seduced by the siren song of AI.

The problem is that AI is a consumer of data, not a creator of it. If your portfolio servicing platform is a walled garden where data goes to die, your AI is just going to be a very expensive way to hallucinate about your delinquency rates.

To make an “assistant” or an “agent” work, it needs to be able to reach into the live portfolio-the contracts, the collateral records, the in-life adjustments-and pull out the truth.

This requires an MCP (Model Context Protocol) server approach, where the AI is given a governed, secure pipe into the actual records of the business. But you can’t build that pipe if you’re still manually typing bank files into a ledger. The “innovation” is the pipe, not the AI at the end of it.

I once spent three days mediating a conflict between a Head of Operations and a Chief Technology Officer at a large finance house. The Ops leader was furious because a major system upgrade had been delayed in favor of a “customer-facing AI pilot.”

The CTO’s response was telling: “The board doesn’t understand ‘back-end reconciliation.’ They understand ‘digital transformation.’ I have to give them what they can see.”

As a mediator, my job-much like Paul K. would suggest-is to find the underlying need that both parties are ignoring. The need isn’t for AI, and the need isn’t even for automation. The need is for integrity.

An organization loses its integrity when the story it tells the world about its technology diverges too far from the daily experience of the people using it.

When Dana spends six hours matching payments, she is absorbing the “technical debt” of the company. She is the human buffer for a broken system. The irony is that the more efficient Dana is, the less likely the company is to fix the problem. Her competence masks the system’s failure.

We see this in every sector of commercial finance. We prioritize the “origination” side-the shiny front-end portals where the money comes in-and we neglect the “servicing” side where the relationship actually lives.

Equipment finance is an in-life business. A lease isn’t a transaction; it’s a multi-year conversation involving tax changes, collateral swaps, and mid-term restructures. If your servicing platform isn’t API-first, you aren’t just slow; you’re opaque.

The real innovation isn’t a chatbot that tells a customer their balance. The real innovation is a platform that ensures the balance is correct in real-time, automatically, without Dana having to touch it. Once you have that-once you have a single source of truth that is accessible via API-then the AI becomes almost trivial to implement. It becomes a tool rather than a totem.

I finally got back into my car today. A locksmith showed up with a thin piece of metal and a wedge. It took him forty-five seconds. I felt like an idiot for spending forty minutes staring at the keys through the glass.

But that’s the thing about being “locked out.” You can see exactly what you need, but you can’t reach it because there’s a barrier in the way that wasn’t there yesterday.

Manual work is the glass. The AI is the car. The data is the keys. Most companies are standing in the parking lot, admiring the car’s sleek lines and the “AI-Powered” bumper sticker, while the keys are sitting on the driver’s seat, completely inaccessible.

If you want to move the car, you don’t need a workshop on the future of transportation. You need a locksmith. You need to break the barrier between your bank files and your ledger. You need to stop asking Dana to be a human API and actually give her an API.

We have to stop treating the back office as a cost center to be minimized and start treating it as the data center of the enterprise. Every manual workaround is a leak in your data integrity.

Every time a servicing manager has to explain why a payment hasn’t posted yet, your “innovation” score drops, regardless of what’s on your roadmap.

True modernization is quiet. It doesn’t usually happen in workshops with facilitators and sticky notes. It happens when a servicing manager realizes she hasn’t seen Dana with a lockbox spreadsheet in three weeks because the system is finally doing its job.

It happens when the “invisible” work becomes so efficient that it actually disappears.

Until we value the invisible as much as the visible, we aren’t innovating.

We’re just decorating the waiting room of a business that’s still running on paper and grit. The keys are right there on the seat. We just have to decide if we’re going to keep staring through the glass or if we’re finally going to open the door.