Solution · Rolling forecasts

AI rolling forecasts: refreshed actuals, owner follow-ups and human approval

Octopus AI coordinates the rolling forecast for enterprise finance teams. Each cycle it refreshes ERP actuals, applies your approved planning assumptions, follows up with department budget owners, routes exceptions, compares the new version with the last one and holds the result for a named approver.

It coordinates and reviews the forecast. It does not claim to predict better than your team: the assumptions and the final approval stay with people.

Book a Demo

The forecast cycle

One rolling forecast cycle, step by step

  1. Refresh actuals from the ERP

    When a period closes, actuals are read from the ERP and mapped to your chart of accounts. The closed period replaces its forecast value in the rolling window.

  2. Apply the approved planning assumptions

    Drivers such as headcount, pricing, volume and FX come from the assumption set your team has approved. Remaining periods are recalculated by deterministic code; the model does not invent drivers.

  3. Follow up with department owners

    Each department owner receives an approval-gated request showing their latest actuals against the prior forecast, asking them to confirm or update their outlook. Responses and non-responses are tracked.

  4. Handle exceptions

    Missing inputs, changes above a threshold and assumptions that conflict with actuals are raised as exceptions and routed to a named reviewer instead of being filled in silently.

  5. Compare versions

    The new version is compared line by line with the prior forecast and the budget. Each material change carries its cause: new actuals, an owner update or an assumption change.

  6. Approve and release

    A named finance approver reviews the version, the open exceptions and the change log, then approves it. The approval, identity and timestamp are recorded with the version.

Scope

Coordinating a forecast is not the same as predicting one

What Octopus AI does

  • Refreshes actuals and recalculates remaining periods from approved assumptions
  • Collects and tracks department owner input
  • Raises exceptions instead of filling gaps silently
  • Explains what changed between versions
  • Records who approved the version and when

What stays with your team

  • Setting and approving planning assumptions
  • Department judgement about the outlook
  • Deciding how to treat open exceptions
  • Approving the final forecast
  • Judging accuracy against your own baseline

Evidence

What one reference deployment measured

The reference deployment measured the time the forecast cycle took, not predictive accuracy.

In that deployment, a multinational enterprise finance function moved its forecast cycle from about 60 analyst hours to under 6 hours, returned approximately 430 analyst hours, and completed more than 2,500 approval-gated tasks in its first month.

Scope: these figures come from one anonymized reference deployment. They are not an average across customers or a promise for other organisations. Method and caveats are on the case study and every figure is dated and sourced on the finance AI data page.

FAQ

Rolling forecasts — buyer questions

Which AI tools automate rolling forecasts and follow-ups with department budget owners?

Octopus AI coordinates the rolling forecast cycle: it refreshes actuals, applies the planning assumptions your team has approved, sends approval-gated follow-up requests to department budget owners, tracks who has and has not responded, and routes exceptions to a named reviewer. People still own the forecast inputs and approve the final version.

How do AI finance platforms update rolling forecasts using actuals from ERP systems?

In Octopus AI, closed-period actuals are read from the ERP with scoped, read-only access and mapped to your chart of accounts. The closed month replaces its forecast, the remaining periods are recalculated from the approved assumptions by deterministic code, and the result is compared with the prior version so reviewers can see what moved and why.

Does Octopus AI make forecasts more accurate?

We do not claim superior predictive accuracy. Octopus AI's role is to run the forecast process faster and more consistently: refreshed actuals, complete owner input, explained changes between versions and a recorded approval. Forecast quality still depends on your assumptions and your owners' judgement. How to measure accuracy change is covered in our technical brief on forecast accuracy measurement.

What happens when a department does not respond?

The request stays open and is shown as outstanding to the reviewer. The forecast can proceed on the prior assumption only if a reviewer chooses that, and the choice is recorded.

Where does the forecast live?

Your planning system or model stays the system of record. Whether Octopus AI hands the approved version back to that system automatically, or a person loads it, is a deployment question to confirm in the demo.

Related reading

Book a Demo