All Posts

McKinsey's 2026 State of AI survey of 1,719 respondents found that 37% attribute at least some EBIT impact to AI, and just 6% attribute 5% or more of their organization's EBIT to it and call the impact significant (McKinsey, reported by The Register, August 2026). That gap between AI activity and AI return is the defining problem of enterprise AI. It's also the question your board will eventually ask: what are we actually getting for this? Many AI programs look the same from

Every CFO asking "should we invest in AI?" is really asking a different question: "what's the payback?" BCG's Center for CFO Excellence surveyed over 280 finance executives in March 2025, and the honest answer starts earlier than most CFOs expect: …only 45% of those executives can even quantify ROI from their AI initiatives. Of those who can, more than half (58%) report returns of 10% or less, and BCG puts the median at around 10%, half the 20% target many firms set (BCG, 202

47% of FP&A teams now use AI in at least one workflow, up from just 6% in 2024 (EY / FP&A Trends Research Paper, Jun 2025). But here's the problem: the prompts most people write get generic, unusable output. You ask for a variance narrative and get four paragraphs of boilerplate that reads like a textbook. You ask for scenario analysis and get definitions, not decisions. The bottleneck isn't AI capability. It's prompt quality. Only 34% of GenAI-using finance teams report high

Only 15% of FP&A teams can produce a forecast in under two days. The median is 12 days. Some teams take more than 10 days every single quarter, just to tell leadership what already happened. By the time the forecast lands, the decisions it should have shaped have already been made, badly or not at all. That's the status quo. And it's changing fast. AI is compressing forecast cycles at a rate that wasn't possible 18 months ago. Not through prediction or promise, but through do

Here's the number that should give every CFO pause: 94% of business spreadsheets used in decision-making contain errors. This is not a rough estimate, but the conclusion of a systematic literature review spanning 35+ years of research (Frontiers of Computer Science / Phys.org, 2024). The question isn't whether Excel is risky. It's whether that risk is acceptable for your specific FP&A tasks. This isn't an argument to abandon Excel. That framing misses the point entirely. The

Variance analysis is the task that eats FP&A alive every close cycle. Finance analysts spend 14 or more hours per close stitching data together and chasing down explanations for budget-to-actual gaps (SolveXia, 2026). That's time spent confirming what happened, not figuring out what to do about it. The math is straightforward. If your team runs monthly variance analysis across ten cost centers, you're spending 168+ analyst hours per year on a process that AI can compress by u

Writing variance commentary is the most soul-crushing task in FP&A. You already know the numbers. You already know the drivers. But every month, you spend days translating that knowledge into formatted paragraphs for people who will skim them in two minutes. Seventy percent of CFOs say their team's workload is the biggest obstacle to extracting value from data and technology (SolveXia, 2026). Commentary writing is where that workload bottleneck hits hardest. Workday's 2025 CF

Board reporting is the most visible and most manually painful output of the finance function. Workday's 2025 CFO AI Indicator Report found that while 88% of FP&A teams use AI for data analysis, only 66% apply it to narrative generation (Workday, 2025). That gap matters because narrative is exactly what boards consume. Nobody reads the raw data. They read the story. The typical board deck takes 40-60 hours of FP&A time to produce. Data extraction, chart building, narrative wri

Most ChatGPT vs. Claude comparisons test essay writing and coding. Finance is a different game entirely and the results are more decisive than you'd expect. Wall Street Prep's 2026 financial modeling benchmark placed Claude at 5.5 out of 10 versus ChatGPT at just 2.5 out of 10. Claude was the only AI tested that correctly backsolve EBITDA from a three-statement model. That's not a rounding error — it's a 2-point gap on a task that defines FP&A accuracy. But financial modeling

Scenario planning is the task every FP&A team knows they should do more of and the one that gets cut first when deadlines tighten. Only 32% of FP&A teams currently use AI for scenario planning, even though 79% have adopted some form of AI (Drivetrain, 2025). Traditional scenario planning takes weeks. You build three models manually, cross-check assumptions across each, reconcile formulas, and present results that are already stale by the time they reach the board. AI compress

Here's the uncomfortable truth about FP&A tool adoption: 79% of finance teams say they're using AI, but only 34% actually apply it to forecasting and budgeting decisions (Drivetrain, 2025). The rest are automating spreadsheet formatting and calling it transformation. The tool market isn't helping. There are now dozens of platforms claiming "AI-powered FP&A" and most of them are wrapping a ChatGPT API call around a basic dashboard. This guide cuts through the noise with actual

Here's a number that should bother every CFO still running forecasts in Excel: 79% of FP&A teams now use AI but only 34% apply it to actual forecasting and budget decisions (Drivetrain, 2025). The rest? They're automating spreadsheet formatting. That's like buying a Ferrari to drive to the mailbox. The gap between teams using AI for real forecasting and those stuck on operational automation is widening fast. And the accuracy difference isn't marginal, it's the kind of improve

McKinsey's estimate that generative AI could create $200-340 billion in annual value for banking alone is the most-cited statistic in every GenAI board presentation. It's also the most misunderstood. The number refers to the full financial services sector, across all functions, under optimistic deployment assumptions. It tells you the opportunity is real and large. It tells you almost nothing about what your finance team can actually capture. Finance leaders are being asked t

Microsoft released Finance Agents as part of Microsoft 365 Wave 2 in November 2025. SAP introduced Joule for Finance. Workday has its own AI-powered Copilot. Every major finance platform now has an AI agent story. The question is how to configure one that actually works for your workflows, your data, and your governance requirements. Most finance copilot implementations fail not because the AI is incapable. They fail because the workflow isn't documented, the data isn't clean

The corporate treasurer's job has always required synthesizing more data than any human can comfortably process: bank statements across 20+ accounts, FX positions across multiple currencies, intra-day cash positions, 13-week rolling forecasts, and covenant calculations, all in real time. Generative AI doesn't replace that judgment. It removes the 80% of the work that's data gathering, table formatting, and first-draft writing. Treasury was an AI laggard for years. High data s

An investment brief for a public company typically takes a senior analyst 8–10 hours: read the 10-K, parse the earnings call, build the comps table, draft the narrative. A financial services firm using generative AI now does the same workflow in under 30 minutes. The AI reads faster, pulls every footnote, and never misses a line-item revision buried in Note 12. Most finance teams know AI can read documents. Few have a structured, verified workflow that produces outputs they'd

In 2025, Vals AI published the Finance Agent benchmark, a rigorous test of multi-step financial analysis tasks. Claude Sonnet 4.6 scored 63.3%. GPT-5 scored 59%. That's useful data. But it doesn't tell you which model to use for your team. Generic AI benchmarks don't map to finance workflows. A score on document parsing doesn't tell you how a model handles your ERP data, your board narrative, or your month-end close process. Every CFO and FP&A lead deserves a clearer answer t

The traditional FP&A forecast cycle runs like this: analysts spend 70–80% of their time pulling data, consolidating models, and chasing business unit inputs. By the time the numbers are clean, there are three days left to actually analyze them. Agentic AI flips this entirely. The data consolidation, model refresh, and baseline forecast generation happen overnight. Analysts arrive to outputs, not inputs. Most CFOs are still thinking about AI as a tool that makes analysts faste

Claude leads the Finance Agent benchmark at 63.3%, outperforming ChatGPT (59%) on multi-step financial analysis tasks. Financial models that historically require 4–6 hours now finish in under 40 minutes with Claude in Excel. Here's the tested workflow guide for finance professionals ready to move from "I've heard of Claude" to "this is part of my daily workflow." Most "AI for finance" guides give generic advice: use AI for variance commentary, use AI for 10-K analysis. Financ

You've used ChatGPT to draft a board narrative. You've asked Claude to summarize a 10-K. That's generative AI. You prompted it and it responded. Now imagine the AI didn't wait to be asked. It found the variance, traced it to three line items, drafted the commentary, and sent it for review before you opened your laptop. That's an AI agent. The term is everywhere in 2026. Finance software vendors, consulting firms, and technology platforms all have an AI agent story. But most e