How to Use AI for Financial Forecasting

Forecasting is one of the most time-consuming things a finance team does. It’s also one of the areas where AI can genuinely change how long it takes — if you know where to apply it.

This isn’t about AI replacing your forecast model. It’s about removing the friction from the process — the parts that eat time without requiring your judgment — so you can spend more of your week on the analysis and stakeholder work that actually matters.

Where forecasting time actually goes

Before you can use AI well in forecasting, it helps to be clear about which parts of the process are genuinely slow. In most finance teams, it’s usually some combination of:

• Chasing budget holders for updated estimates and assumptions
• Consolidating inputs from multiple sources into a single model
• Writing the narrative to accompany the numbers
• Preparing the output in a format suitable for different audiences
• Answering the same questions about the forecast from different stakeholders

The actual forecast logic — the model, the assumptions, the judgment calls — is often a smaller share of the time than the process around it. That’s where AI helps most.

Drafting forecast commentary

Give an AI model your forecast movements and ask it to draft the narrative. Paste in the key variances versus prior forecast or budget, describe any known drivers, and tell it the audience. You’ll have a structured first draft in under a minute.

The draft won’t include the specific insight you have about why a particular project has slipped or why headcount is coming in lower than planned. That’s your edit to make. But having a structured starting point — rather than a blank page under time pressure — cuts writing time significantly.

Use this prompt as a starting point:

Write forecast commentary for [period] for [business area].
Audience: [CFO / Finance Director / budget holder].

Forecast vs prior: [£X] [favourable/adverse]. Key movements:
- [Item]: [£X] movement. Reason: [brief explanation]
- [Item]: [£X] movement. Reason: [brief explanation]

Full year outlook: [on track / revised to £X / risk of £X].

Write 2-3 paragraphs: current position, key drivers, full year view.

Challenging assumptions

One of the most valuable things you can do with AI in forecasting is use it to pressure-test your own assumptions before a stakeholder review. Paste in your key forecast assumptions and ask the model to identify the weakest ones and suggest the questions a CFO or Finance Director would likely ask.

This is useful not because the AI knows your business — it doesn’t — but because it can apply general financial logic to flag where assumptions look inconsistent or optimistic. It’s the equivalent of a sense-check from a colleague who’s read a lot of finance commentary and knows how these conversations usually go.

Excel modelling support

If your forecast model is in Excel (and most are), AI can help with the mechanics. Formula errors, VBA automation for consolidation, Power Query for pulling in data from multiple sources, dynamic range references that update as you add rows — describe what you need and you’ll get working code to test.

For teams that run a rolling forecast — updating 12 or 18 months forward each month — the setup and maintenance of that model can be time-consuming. AI won’t redesign your model for you, but it can help you build automation that removes repetitive manual steps from the monthly update cycle.

Scenario analysis

When you need to present a range of outcomes rather than a single point forecast, AI can help structure the scenario logic. Describe your base case and ask it to help you define what the upside and downside scenarios look like — which assumptions change, by how much, and what the financial impact would be.

This is particularly useful when you’re new to an area or working on a forecast for a business unit you’re less familiar with. The AI can help you think through the drivers systematically before you build the numbers.

What AI doesn’t change

The quality of a forecast depends on the quality of the assumptions behind it, and assumptions come from understanding the business. AI doesn’t know that your major IT programme is six months behind schedule, that the planned headcount increases have been quietly shelved, or that the new product launch is at risk. You do.

AI helps with the process around the forecast. The judgment inside the forecast is still yours — and that’s the part that matters most for the organisation you’re partnering with.