AI In Finance: Glenn Hopper On Bottom-Up Adoption

I see it every day in the classrooms and the boardrooms. Finance teams are no longer waiting for a memo from the top to start using new tools. They are already running experiments on their own laptops.

In the hallways of the world’s biggest banks, people are using artificial intelligence to write reports and check for errors before their bosses even ask for a strategy.

It is a quiet takeover from the bottom up. This is a big shift for a field that loves rules and slow changes.

Now, the people at the top are racing to catch up with their own staff.

They have to find a way to make these tools safe without stopping the work that has already started.

Precision is the old goal, but speed is the new reality.

This shift toward rapid, bottom-up adoption was a central theme at recent Oracle NetSuite events, where the talk has moved away from saving money. Most people think AI is about cutting jobs or lowering costs, but they are wrong.

The real driver is how easy it is to plug these tools into what people already do. For example, the Model Context Protocol (MCP) is now a big deal because it lets different systems talk to each other without a fuss. It makes the technology vanish into the background.

When the tool is invisible, people use it more. It is like electricity; you do not think about the wires, you just want the light.

If a tool is hard to use, it stays on the shelf regardless of its price.

Efficiency is a side effect, not the main target.

While integration is becoming easier, a significant challenge remains regarding the human element. Glenn Hopper from VAi Consulting says the real gap is between what people know about finance and what they know about tech. You can have the best computer in the world, but it is useless if the person using it does not understand how it thinks.

Some bosses are so scared of mistakes that they lock everything down. This is a mistake.

When you make the rules too tight, employees find sneaky ways to use AI anyway.

That is how you lose control and skip the audit trail.

You need humans who can argue with a machine and win.

The Stress Test Of Modern Oversight

The friction between clandestine AI use and rigid corporate rules means the old ways of checking work are breaking. Because workers started using AI before there was a plan, the data is often scattered in places where it should not be. To pass this test, companies must build systems that show every step the machine took. Transparency is the only way to stay out of trouble with the law. If you cannot explain why the AI flagged a trade as fraud, you cannot use it. Trust is earned through clear steps, not through magic black boxes.

Most firms are still failing this test because they care more about the output than the process.

Winning The Tech Integration Game

To move beyond simple oversight and achieve true productivity, winning now looks different than it did two years ago. It starts with making AI a part of the air you breathe. Instead of a separate app, the tech is now inside the ledger and the contract sheet. By using systems that share data easily, finance teams spend less time fixing old numbers.

They can look at the future instead.

This means the job of a bean counter is dying.

In its place, we see the rise of the finance architect.

These are people who build paths for data to flow. They do not just report the weather; they try to change it. Putting the tech in the right spot is more important than having the newest model.

Why The CFO Is Now A Coder

The evolution of the finance architect is supported by data from the Stanford 2026 AI Index. It shows that AI is moving faster than our ability to teach it. My analysis shows a weird trend: the best finance leaders today look more like software engineers. For instance, they are using "Agentic AI" to handle multi-step tasks like monthly closings without human help. This connects the dots between raw data and real choices.

If the AI can draft the narrative for a quarterly report, the human must become a high-level editor.

We are moving from "doing the work" to "guiding the work." This is a radical change in what we teach in business school.

We used to teach kids how to use a calculator; now we teach them how to manage a digital workforce.

You are no longer managing people; you are managing a mix of humans and code. It is a bit like being a chef who has to watch ten automatic ovens at once.

How To Run A Modern Finance Desk

To successfully manage this mix of human talent and automated logic, you must follow a clear method. First, stop buying big AI platforms that do not talk to your current software. Use small, open tools that connect via APIs. Second, give your team a "sandbox" where they can play with data without breaking the main system.

This keeps them from using private accounts to get work done. Third, set up a weekly "model check" where humans review the AI's logic.

This is not about being mean to the machine; it is about keeping it honest.

Fourth, hire people who are curious about how things work. You need people who ask "why" more than they ask "how much." Finally, make sure every piece of AI work has a human name attached to it. Accountability cannot be outsourced to a chip. If the machine makes a mess, a person still has to clean it up.

Big Questions For The New Era

As these methods become standard, we must confront the long-term implications of a digital-first desk. How do we value a company when its main workers are digital agents? What happens to the entry-level job when AI does all the basic tasks? Can we ever truly trust a machine to be ethical in a market crash? These are the things that keep me up at night. To find some answers, look into these areas:

• The 2026 update on Machine Learning Governance and its impact on global banks.
• Research from OpenAI regarding their latest "Reasoning" models and how they handle math.
• Case studies on "Ambient Finance" where AI runs in the background of everyday transactions.
• The latest white papers on "Interoperability Standards" for financial data sharing.