Category: AI in Finance
-
When AI Financial Forecasts Are Wrong: A Skeptic’s Framework for Sanity-Checking
AI-generated forecasts fail in specific, predictable ways. I have watched three major vendor tools (Anaplan, Pigment, Mosaic) produce revenue forecasts that were within 3% of actuals for six consecutive months, and then miss the seventh by 22% on the same P&L line. Not because the tool broke. Because the input…
-
The Board Deck AI Prompt That Cuts Prep From 8 Hours to 45 Minutes
Board decks are the most-time and least-fun deliverable in the CFO calendar. At the $180M SaaS company where I ran finance last year, I spent between 7 and 9 hours on the CFO-narrative sections of a monthly board pack. Not the analyst tables. Not the schedules. Just the narrative, the…
-
Why Most CFOs Are Using AI Wrong (And What The Top 10% Do Differently)
KPMG’s 2026 Global AI in Finance survey found active use of AI in the finance function has more than doubled since 2024, rising to 75%. But adoption and outcomes are two different things. In my own working conversations with middle-market and PE-backed CFOs over the last 18 months, only about…
-
How to Structure Your Chart of Accounts for AI-Powered FP&A
Half the FP&A pain I see in middle-market finance orgs traces back to a chart of accounts built for 2014, not 2026. In middle-market finance orgs I have worked with, the pattern is remarkably consistent: the companies with the most accurate forecasts run leaner charts, often less than half the…
-
The Weekly AI-Audit Every Controller Should Run (10 Prompts + Templates)
I ran this weekly audit at a $180M SaaS business last quarter. It caught two revenue recognition errors that would have blown a Q3 covenant test, one duplicate vendor payment ($47,200 to a marketing agency we paid twice in 11 days), and a stock comp accrual that was $310,000 light…
-
6 Things You Should NEVER Ask an LLM as a CFO
TL;DR Every CFO should have a mental “do not ask” list for LLMs. It is not about being paranoid. It is about knowing where the model is confidently wrong or where the tool is a legal liability. Below are six categories that have burned people I know, with the failure…
-
The AI-Assisted CFO Weekly Cadence: Monday Through Friday
TL;DR A finance week has five distinct modes: review, prep, pulse, deep work, and wrap. AI helps most when you match the tool to the mode instead of trying to “use AI more” as a general goal. Below is a Monday-through-Friday operating rhythm with the specific prompt to run each…
-
AI-Generated Investor Updates: A Templated Prompt That Sponsors Actually Respect
TL;DR Sponsors read the first paragraph and the asks. Everything else is skimmed. Below is the exact monthly update prompt I have used to generate first drafts that partners forwarded to their IC without meaningful edits. The prompt is 90% of the value. The edit pass is where you strip…
-
Custom GPT vs Claude Project vs Copilot Agent: Which Fits Your Finance Workflow
TL;DR Claude Projects win for long-form reasoning over a big file set (QoE, budget narratives, board writeups). Custom GPTs win when you need a shareable, single-purpose tool for a repeat workflow (invoice coding, expense triage). Copilot Agents win when your data lives in Excel, SharePoint, and Outlook and you cannot…
-
Build Your Board Deck with AI: The 6-Prompt Playbook
TL;DR Six prompts. Copy them in order. Feed a P&L export, a cash file, a Slack digest, and a plan. You get a board-ready deck draft in about 45 minutes instead of two days. You still edit the numbers, own the narrative, and cut the caveats. Everything below is what…
-
The CFO’s Guide to Uploading Financials to LLMs Without Privacy Risk
Three redaction techniques, four enterprise LLM tiers, a safe-vs-unsafe table, and ten vendor questions. The full playbook for using AI on real financials without a compliance incident.
-
Building a Rolling 13-Week Cash Flow With Claude
The 4-prompt sequence for a rolling 13-week direct cash forecast. Input structure, expected outputs, weekly re-baseline routine, and the verification checkpoints that catch Claude’s arithmetic errors.