All Posts

The CFO job has not been reinvented by AI. It has been reshaped in specific ways that are worth naming, because the reshape is happening whether finance leaders plan for it or not. Three shifts are already visible. The first is that production work is getting cheaper faster than judgment work. Variance commentary, forecast memos, board narrative, audit pre-reads — all of these are being compressed by AI. The pieces of the CFO job that were about production are taking less tim

AI is good at a lot of finance work. It is also wrong in specific, predictable ways that every CFO should know before they build it into their workflow. Three failure modes show up consistently. The first is confident wrongness on numbers. Ask an AI model to calculate a compound growth rate, a variance percentage, or a ratio, and it will give you an answer. That answer is often wrong. Not directionally wrong. Specifically wrong, in a way that looks right on the page. This is

Variance commentary is one of the highest-volume writing tasks in finance. Every month. Every quarter. Every forecast cycle. Every function, every region, every P&L line. It is also one of the easiest tasks to accelerate with AI, if the prompt is built correctly. Most finance teams prompt AI for variance commentary the generic way. "Write a variance commentary for the following numbers." The output is directionally fine and operationally useless. Too generic. No structure. No

Board prep eats weekends. That has been true for as long as I have been a CFO, and for most finance leaders it is still true now. AI does not make board prep easier. It compresses the parts of board prep that do not require judgment, so the finance leader can spend more time on the parts that do. The workflow I use has four stages. Stage one is the narrative frame. Two to three weeks out, I use AI to help structure the argument the deck is going to make. Not to write content.