Skip to content

Forecast updates

Your forecast should change when the business does.

Our FP&A Agent identifies changes since the last forecast, investigates their financial impact and proposes updates to the model your finance team already uses.

Your model stays yours. The agent keeps it informed.

See a forecast updateTalk to us
Illustration: the FP&A Agent working against a finance team's own forecast model, FY26 Forecast Model.xlsx, version 12, last updated 1 September; the company's fiscal year is the calendar year, so Q4 FY26 is October to December 2026. Asked whether anything since version 12 changes Q4, it lists four changes: hiring that ran ahead of the headcount plan, recurring, +$1.8M in Q4, from payroll journals in SAP S/4HANA and Headcount Plan.xlsx; a new committed cloud infrastructure spend, recurring, +$1.1M, from the amended contract and SAP S/4HANA; the annual conference, a one-off Q3 overspend, $0 in Q4; and a customer renewal expected in September that has slipped to October with no signed date, uncertain, so no change is proposed yet, from Salesforce and the account owner. It proposes Q4 updates: payroll from $22.6M to $24.4M, cloud infrastructure from $8.5M to $9.6M, and marketing unchanged at $6.3M, a net +$2.9M. Status: needs review by Megan Carter, FP&A Manager. Nothing in the model changes until she accepts.

The forecast starts getting stale as soon as you finish it.

Version 12 went out on 1 September. By the end of the month the business had changed four times and Q3 had closed, and the model still said what it said on day one.

Illustration: September at the company, as a timeline of date, event and classification. 1 September: forecast version 12 published, Q4 operating expenses at $46.3M. 4 September: payroll differs, with 22 people, and the merit increase, paid ahead of the dates the model assumes; recurring. 11 September: cloud contract amended, with new committed spend from October; recurring. 16 September: a large renewal slips from September to October with no signed date; uncertain, so the agent proposes no change and asks the account owner. 21 September: the annual conference closes $0.9M over budget; one-off. 30 September: Q3 ends with operating expenses forecast at $46.2M against a $42.0M budget, a $4.2M variance to investigate. Through all of it the model stays at version 12.

Traditional process

  1. Wait for the October cycle
  2. Investigate every change by hand
  3. The forecast catches up

Weeks after the business moved

With our FP&A Agent

  1. A change happens
  2. Detected
  3. Investigated
  4. Quantified
  5. Proposed

While the change is still news, with its evidence

Planning software can calculate the forecast. Someone still has to work out what changed.

The agent watches what changed.

One change, followed through. Each stage takes the last one's output as its input, so the proposal arrives with its working.

Illustration: one change followed through six stages. Detect: payroll shows 22 people, and the merit increase, ahead of the dates the model plans. Investigate: 9 account executives started on 22 July, planned for 15 December, and 13 more, with the merit increase paid from July and planned for January. Classify: recurring, because they stay on payroll. Quantify: +$1.8M in Q4 FY26, with the model's own logic. Propose: payroll from $22.6M to $24.4M. Review: needs review by Megan Carter, FP&A Manager.

Changes that should reach the forecast sooner.

Each kind of change shows up somewhere before it shows up in the forecast. The agent reads it there and names the line it should move.

Illustration: six kinds of change, where the agent sees each and the forecast line it moves. Headcount, from payroll journals and the headcount roster, moves payroll. Revenue, from Salesforce opportunities and billing, moves revenue. Vendor commitments, from contracts and AP invoices, move committed spend. Operating expenses, from the general ledger, move department expenses. Cloud and usage costs, from cloud invoices and usage reports, move cloud infrastructure. Cash, from bank activity and AR ageing, moves the cash plan.

Worked example

The variance that should change Q4.

Q3 is ending $4.2M over budget. The question for the forecast isn't why. It's how much of it is still there in October.

Illustration: Q3 FY26 operating expenses, July to September 2026, $4.2M over a $42.0M budget, taken from variance to forecast. The five drivers are headcount +$2.1M, cloud infrastructure +$1.4M, the annual conference +$0.9M, professional services −$0.5M and other +$0.3M. The agent classifies what carries into Q4: headcount +$1.8M, cloud infrastructure +$1.1M, the conference nothing, professional services −$0.3M and other +$0.3M, $2.9M in all, while $1.3M does not carry. Services and other are already in forecast version 12, so it proposes two updates: payroll +$1.8M and cloud infrastructure +$1.1M, with marketing unchanged because the conference was one-off.

Variance analysis becomes an input to the forecast, not a separate spreadsheet exercise.

Keep the planning system. Remove the manual investigation around it.

The model stays where it is, with its formulas, its assumptions and its owners. The agent does the work of finding out what the model should know.

Your planning software

  • Stores the model
  • Maintains assumptions
  • Runs forecast calculations
  • Supports planning workflows

Our FP&A Agent

  • Finds what changed
  • Investigates why
  • Classifies the change
  • Quantifies the impact
  • Proposes updates
  • Attaches evidence

We do not replace the planning system as the system of record for the forecast.

Nothing changes until finance accepts it.

Every proposal carries its reason and its evidence, and waits for a named reviewer.

Illustration: one proposed forecast update under review. Payroll for Q4 FY26, October to December 2026, from $22.6M to $24.4M, +$1.8M. The reason: 22 people, and the merit increase, are on payroll ahead of the dates the model assumes: 9 account executives from July planned for mid-December, 5 analytics engineers from August planned for January 2027, 8 support specialists the plan did not include, and the merit increase, paid from July and planned for January 2027. The engineers who started in July were budgeted from 1 September, so they do not move Q4. The evidence is payroll journals in SAP S/4HANA, the headcount roster and Headcount Plan.xlsx. Prepared by the FP&A Agent; reviewer Megan Carter, FP&A Manager; status needs review. She can accept, edit or reject it; accepting approves it for version 13 of the team's own model.
Forecast updatePayroll · Q4 FY26 · Oct–Dec 2026
Current
$22.6M
Proposed
$24.4M
Change
‎+$1.8M

Reason

22 people, and the merit increase, are on payroll ahead of the dates the model assumes.

  • Account executives (9)From 22 Jul · planned 15 Dec
  • Analytics engineers (5)From 4 Aug · planned Jan 2027
  • Support specialists (8)From 3 Aug · not in the plan
  • Merit increaseFrom 1 Jul · planned Jan 2027

The 14 engineers who started in July were budgeted from 1 September, so they don't move Q4.

EvidencePayroll journalsHeadcount rosterHeadcount Plan.xlsx
Prepared by
FP&A Agent
Reviewer
Megan CarterFP&A Manager
Status
Needs review

Waiting on Megan Carter. The model doesn't change until a decision is made.

Approving records the decision against the proposal. FP&A carries it into version 13 of its own model.

The agent proposes. Finance decides.

Every forecast change has a reason.

Each version keeps what changed, why, who proposed it and, once approved, who approved it, so the forecast can be explained a quarter later. Select a version.

Illustration: the forecast model's versions as a timeline; Q4 means Q4 FY26, October to December 2026. Version 10, published 1 August: Q2 actuals rolled forward, Q4 left at its $45.7M budget, proposed by Megan Carter, approved by Sarah Chen. Version 11, published 15 August: sales commissions +$0.6M in Q4 because revenue was running above plan in July, proposed by the FP&A Agent, approved by Sarah Chen; Q4 $46.3M. Version 12, current since 1 September: professional services −$0.3M and other +$0.3M, proposed by the agent, approved by Megan Carter; Q4 still $46.3M. Version 13, proposed on 30 September: payroll +$1.8M and cloud infrastructure +$1.1M with marketing unchanged, proposed by the agent and needing Megan Carter's review; Q4 would be $49.2M. Version 13 is shown open.

v1330 SepStatusNeeds review

What changed
Payroll +$1.8M · Cloud infrastructure +$1.1M · Marketing unchanged
Why
Hiring ahead of plan and an amended cloud commitment; the conference was one-off
Proposed by
FP&A Agent
Reviewer
Megan Carter
Q4 impact
+$2.9M
Q4 operating expenses
$49.2M

Questions

Where our agents start and stop.

  • Do you replace our planning software?

    No. Your planning system or model stays the system of record for the forecast. Our agents work out what should change in it.

  • Do you generate forecasts using AI?

    No. Our agents identify business changes, investigate their impact and propose updates. The forecast itself comes from your model, its assumptions and its logic.

  • Can your agents update the forecast automatically?

    No. They propose each change with its reason and evidence. FP&A accepts, edits or rejects it, and carries what it accepts into its own model.

  • How do your agents know something changed?

    They compare the assumptions in your forecast with new information in your connected systems: payroll journals, contracts and invoices in the ERP, the pipeline in the CRM.

  • Can we see why a forecast assumption changed?

    Yes. Each proposal keeps its reason, its evidence, who reviewed it and the version of the model it applies to.

The productFP&AInvestigate what changed, carry it into the outlook and model what happens next without rebuilding the context.Explore FP&A

Bring your current forecast.

We'll show you which changes the agent would investigate and what updates it would propose.

Book a working session