9 min read
Your CFO is good at their job.
I want to start there because everything that follows will sound like an attack on finance, and it is not. Your CFO manages capital allocation, risk exposure, and fiduciary responsibility for a complex organization. That is genuinely difficult work. Most people could not do it.
Your procurement team is also good at their job. They negotiate contracts, manage vendor relationships, and protect the company from bad deals. That matters. It is necessary.
The problem is not that these people are involved in AI decisions.
The problem is that in many companies they are effectively making AI decisions. And they are not qualified to do that. Not because they are not smart. Because they have not done the work of understanding what AI actually changes inside a business.
That distinction matters enormously right now.
The Printer Problem
Here is what I see in company after company.
Someone in engineering or operations wants to adopt an AI tool that changes how their team works. Maybe it writes code. Maybe it triages customer issues. Maybe it drafts contracts or analyzes financial data or automates a workflow that currently takes three people and a spreadsheet.
The request goes to procurement. Procurement routes it to finance. Finance evaluates it the same way they evaluate everything — cost, vendor risk, contract terms, budget line item.
Then someone approves it or rejects it.
And the process looks almost identical to what would happen if someone requested a new printer.
That is the problem.
When a new printer gets installed, someone files a ticket. IT plugs it in. Everyone moves on.
A printer does not change how your analysts think. A printer does not restructure what your customer service team is capable of. A printer does not determine whether your engineering organization ships in two weeks or two months.
AI does all of those things.
And right now, in a lot of organizations, the decision about whether to adopt it — and which version, and for whom, and how — is being made by the same process that handles the printer.
Benjamin Franklin Picking an Airplane Manufacturer
I am not being dramatic when I say your CFO is not qualified to evaluate AI tools.
Think about it this way.
Benjamin Franklin was one of the most brilliant people who ever lived. Inventor. Diplomat. Scientist. Economist. A genuine polymath who understood electricity, governance, publishing, and human nature at a level most people never approach in any single discipline.
Now imagine asking Benjamin Franklin to choose between Boeing and Airbus for a new fleet of commercial aircraft.
He is brilliant. He is also completely uneducated on airframes, turbine metallurgy, avionics integration, and FAA certification requirements. His intelligence is not the issue. His domain knowledge is. You would not hand him that decision and say “you are smart, figure it out.” You would bring in people who understand aviation and let Franklin manage the economics.
That is exactly the situation your CFO is in right now.
They understand capital. They understand risk. They understand contracts and cash flow and budget cycles. Those are real skills. But evaluating what an AI coding assistant actually does to engineering velocity? Understanding which large language model handles regulatory text well enough to trust in a compliance workflow? Knowing whether an agent framework will actually integrate with your existing systems or just demo well?
That is a different kind of knowledge. And your CFO does not have it. Neither does your procurement team. Neither does the committee that approved the last SaaS renewal.
That is not a criticism. That is just honest.
This Is Not About Taking Power Away From Finance
I want to be clear about something because I have watched this conversation go sideways in too many leadership teams.
Your CFO should control the money. Full stop.
Your procurement team should manage the buying process. Full stop.
Finance should set budgets. Procurement should negotiate contracts. Legal should protect the company. Security should enforce guardrails.
None of that changes.
What changes is the sequence.
Right now, in most companies, finance and procurement evaluate the tool first. They narrow the field based on cost, vendor size, contract terms, and risk profile. Then, somewhere near the end, the people who actually understand the work get invited to comment on a decision that has already been mostly made.
That sequence made sense when you were buying software that sat on top of existing workflows. CRM. ERP. Collaboration tools. Expense management. You could evaluate those primarily on commercial terms because they were not going to fundamentally change what your people do every day.
AI is different.
AI changes the work itself. It changes what people are capable of. It changes how many people you need and what skills those people require. It changes which functions can be automated, which roles become more valuable, and which processes can be eliminated entirely.
That is not a purchasing decision. That is an operating decision. And operating decisions require operating knowledge.
What Your CFO Does Not Know Yet
I say this with genuine kindness because I have been in rooms with CFOs who are trying very hard and doing the best they can with the information they have.
Here is what most CFOs do not know about AI right now.
They do not know that the difference between a $20 per seat AI coding assistant and a $60 per seat AI coding assistant can represent a 3x difference in engineering output — not because the expensive one is better at everything, but because it handles the specific patterns your codebase requires.
They do not know that a “free” open-source model running on your infrastructure can cost more in compute, integration time, and maintenance than a commercial API that bills by usage.
They do not know that two AI tools with almost identical feature lists on a vendor comparison spreadsheet can produce wildly different results depending on the domain, the data, and the workflow they are embedded in.
They do not know that AI capabilities are changing every three to six months in ways that make last quarter’s evaluation obsolete — which means a two-year contract locked in today could be buying yesterday’s capability at tomorrow’s price.
These are not failures of intelligence. They are gaps in domain education. And those gaps are completely reasonable. Your CFO has spent their career getting very good at finance. You would not expect an AI engineer to be good at treasury management. The reverse should not surprise anyone either.
What AI Is Actually About to Do
Here is the part that makes this urgent instead of just interesting.
AI is not going to sit politely on top of your existing organization and make things a little faster. It is going to replace whole categories of work.
Not all at once. Not everywhere. But steadily and structurally in ways that most leadership teams are not preparing for.
Entire workflows that currently require a team of analysts will be handled by an agent and a single person reviewing the output. Processes that take three weeks will take three hours. Roles that exist because the system is too slow or too fragmented will stop existing because the system is no longer slow or fragmented.
That is not a future prediction. That is happening now in the companies that are moving.
And the companies that are not moving — the ones where procurement is still evaluating AI tools the way they evaluate printer contracts — those companies are not just falling behind on technology. They are falling behind on capability. On speed. On what their people can do versus what the competition’s people can do.
The CFO needs to understand that. Not so they can panic. So they can fund the right response. That is what a good CFO does — they fund the things that keep the company competitive. But they cannot fund what they do not understand.
How to Actually Do This
The answer is not to sideline finance. The answer is not to go around procurement. The answer is not to make the CFO feel stupid in a leadership meeting.
The answer is education and sequence.
First, the people who understand the work — engineering leaders, operations leaders, domain experts — evaluate which AI capabilities matter. They assess the tools. They test them against real workflows. They determine what actually changes the output.
Then they bring that assessment to finance with a clear business case. Not “we want this cool tool.” Instead: here is what this capability does, here is the productivity change we measured, here is what it costs, here is the risk, here is the timeline.
Finance evaluates the economics. Procurement structures the deal. Legal reviews the terms. Security sets the boundaries.
Everyone does their job. But the domain assessment happens first, not last.
That is the difference between an AI strategy and a purchasing process that happens to include AI.
The Kindest Thing You Can Do
The leaders reading this who work alongside a CFO or procurement team — I want to say something directly to you.
The kindest thing you can do for your CFO right now is not to protect them from this conversation. It is to have it.
They are going to learn that AI changes the economics of your business. The question is whether they learn it from you, with context and good faith, or whether they learn it from the market, after the competition has already moved.
Your CFO is not the enemy. Your procurement team is not the enemy. They are colleagues doing important work with an incomplete understanding of what AI actually changes. That makes them exactly like most of the leadership population right now.
But someone in the room has to say it. Someone has to explain that this is not a printer. That the evaluation process designed for commodity software does not work for a technology that reshapes what your people can do.
If not you, who?
What This Looks Like When It Works
When this works, the CFO becomes one of the most important allies in the AI transition. Because once they understand the capability shift, they can do something nobody else in the room can do.
They can fund it properly.
They can build the budget model that supports rapid iteration instead of annual purchasing cycles. They can create the financial framework that lets teams experiment with AI tools without a six-month procurement process for every new capability. They can explain to the board why AI spending looks different from traditional IT spending — why it is more like R&D than infrastructure.
A CFO who understands AI is extraordinarily valuable. A CFO who does not understand AI but controls the decisions anyway is an extraordinarily expensive problem.
The difference is education. And education is something every leader in the room can help with.
The Conversation Worth Having
Your CFO is brilliant.
Your procurement team is disciplined.
Neither of them should be deciding which AI capabilities your company adopts. Not because they are not trusted. Because they have not yet learned what there is to know.
Help them learn. Bring them into the work. Show them what the tools actually do, not in a vendor demo, but in your workflow, with your data, producing your results.
That conversation is the beginning of an actual AI strategy.
Everything before it is just purchasing.
