People and money
Many long-standing business systems are shifting out from under us right now. Content marketing started changing as soon as ChatGPT launched, not entirely for the better. Software development is changing for sure, at least if you're technical-ish; coding isn't even the right verb anymore. Companies are optimizing supply chains, using AI to interact with customers, and, at least in theory, changing how decisions get made.
It's a lot of change. But whatever industry you're in, two things keep your company going through all the other shifts: people and money.
CFOs are not a vibey bunch
The best CFOs are great storytellers. Hear me out! A financial model explains the past and predicts the future. Some models feel like journalism. Controversial models read more like science fiction.
A good financial model forces its author to be explicit about everything they believe: which ideas will become products and which will fizzle out, how many widgets people will buy, how many people it takes to build the widgets, the temperamental winds of the market, and a lot of other stuff. In fact, financial models can be so fantastical that companies typically build several of them in an effort to forecast the future: a model with baseline assumptions, a best-case scenario, and a worst-case scenario.
When you stop and think about it, CFOs have been working like Claude Code since way before AI was invented. The CFO's job is to take the vibes of all the leaders around them and turn them into something structured and explicit. CFOs are the anti-vibe.
As you might imagine, the world's most successful CFOs were never really a vibey bunch. For instance, if you look at the college majors originally chosen by CFOs at the current Fortune 100, you find a whole lot of accountants.
The top 100 tech CFOs are even less vibey than everyone else, with only 3% (!) of them having majored in the arts or humanities.
AI enters the chat
Like every other corporate leader, CFOs are trying to reinvent their work with AI. But CFOs don't have a lot of patience for hallucinations. Sure, vibes go against their temperament and training, so they start off skeptical. But CFOs also have to deal with hard compliance requirements and markets that fluctuate every time they sneeze. Honestly, CFOs have good reason for insisting on more than vibes.
Can you imagine trying to explain hallucinatory oopsies to a board of directors, or worse, to the public markets? Yeah, no. The bar for AI use within finance is inherently higher than it is for most other parts of the enterprise because getting it wrong is potentially a company-ending event.
On the other hand, CFOs do love efficiency, and the siren call of AI is awfully compelling on those grounds. What's a vibe-hating CFO to do?
AI, but make it vibeless
Thanks to my experience starting and running Textio, I am legitimately hirable as an early-stage CFO. These are not skills I ever sought to have, but alas, we are where we are. Still, I'm not really a finance expert, so I phoned a friend. Parijat Sarkar is the CEO and Co-Founder of Numos, which makes an "AI teammate for finance." I chatted with him about all this last week when Numos came out of stealth.
In most AI products, end users tolerate a certain amount of hallucination in exchange for efficiency. Most of us are willing to accept several cycles of prompting to make the final output accurate. However, finance users are far less tolerant of these iteration cycles. When I was chatting with Parijat, I was struck by the hard requirement to show the AI's sources, reasoning, and audit trail at every step of the finance workflow. For finance leaders, audit-ability is a non-negotiable because the cost of being wrong is so high. This tracks for me based on my experience, but it started me thinking about other work too.
After all, finance isn't the only discipline where hallucinations can be catastrophic. What about legal? How about HR? The stakes are high in many departments. What is it about finance in particular that causes its leaders to insist on the full audit trail before agreeing to use AI?
It's not an accident that those CFOs weren't liberal arts majors.
Kieran
If you liked this story, why not subscribe to nerd processor and get the back issues? Also, why not learn to tell data stories of your own?
My latest data stories | Build like a founder | nerdprocessor.com