September marks the unofficial start of budgeting season. Over the next three months, finance
teams everywhere will build their plans for 2027—and many of them will repeat the same ritual:
take last year’s numbers, add 5%, argue about the marketing line, and call it done.
There’s a better way, and it starts with acknowledging what a budget is actually for. A budget
isn’t a prediction contest. It’s a resource-allocation tool and an accountability mechanism. Its job
is to force explicit decisions about where money goes and to give you a baseline against which
to measure surprise.

Here’s how AI is changing each stage of the process.
Baseline building. The most tedious part of budgeting is assembling the historical picture: what
did we actually spend, by category, by month, and why? AI-assisted accounting platforms now
do this automatically, with clean categorization and the ability to surface anomalies (“your
software spend grew 40% this year—here are the twelve new subscriptions responsible”). What
used to take a week of pivot tables takes an afternoon.

Driver-based forecasting. Instead of budgeting revenue as a single number pulled from
optimism, modern tools let you model the drivers underneath it—leads, conversion rates,
average order value, churn. Machine learning models can suggest realistic ranges for each
driver based on your history and, in some tools, on anonymized benchmarks from similar
businesses. The output isn’t one number; it’s a range with a most-likely case, which is far more
honest.

Continuous reforecasting. The biggest shift AI enables is treating the budget as a living
document. When actuals flow in automatically, variance analysis happens monthly (or weekly)
without manual effort, and the forecast for the remainder of the year updates itself. The annual
budget becomes the starting point for a rolling forecast rather than a stone tablet

Three practical tips as you head into planning season. First, budget at the level you’ll actually
manage—if no one will ever act on the difference between “office snacks” and “office supplies,”
don’t budget them separately. Second, build the downside case, not just the plan; know now
which expenses you’d cut if revenue came in 15% light. Third, involve the people who own the
numbers. AI can draft a budget, but only your sales lead knows the pipeline is soft.

The goal isn’t a perfect budget. It’s a budget that makes disagreements explicit, surfaces
surprises fast, and gets smarter every month.

Next week: the unglamorous goldmine of accounts payable automation