Cash flow forecasting software that shows the low point, why, and what to do about it
Cash seen 13 weeks ahead, every number traced, so less idle cash sits as buffer.
See one case, screen by screen ↓Cash Flow Forecasting is cash flow forecasting software that builds a 13-week weekly forecast and a 12-month monthly view every Monday from your banks, ledger, payroll, tax and credit agreements. Seven agents forecast each driver, check each other, explain last week's gap and draft funding plans; the treasurer opens any number to its source and approves before the forecast goes to the CFO and the lender.
How to choose AI solutions for back-office and enterprise work →
Updated
Every Monday the cash manager downloads prior-day statements from each bank portal in BAI2, MT940 or camt.053, ties the opening balance, pulls the AR aging and AP open items from the ERP and emails the payroll, tax and capex owners for their numbers. Each item goes into the week it should clear, receipts on top and payments below, and the 13-week cash flow forecast reaches the treasurer on Tuesday or Wednesday. FP&A builds a separate 12-month view from the plan, and the two rarely agree.
The misses come from timing. Receivables are placed on their due date while large customers pay late, a sales order and its invoice are counted twice, a tax instalment or an amortisation payment is forgotten, a bank feed is stale. Last week's gap is explained from memory as "timing". Because nobody fully trusts the number, the minimum cash buffer is set high, and borrowing decisions come late and cost more.
Estimated split for one weekly 13-week forecast for one business, by hand and with the solution, in hours.
Seven agents forecast each driver, check each other and explain last week's gap; the treasurer decides and signs.
Monday 9 November 2026. Kestrel & Vane Distribution, the US distribution business of the Kestrel & Vane Group, opens the week with $22.4M across four banks, a $60.0M asset-based revolver and a $15.0M minimum cash policy. Its 13-week forecast, FC-2026-W46, is due to the lender on Tuesday at 17:00. Here is the morning, screen by screen, in the working solution.
The Monday run starts on schedule. The cash position reader reads four prior-day files — Halberd BAI2, Northmere camt.053, Calloway MT940, Pinecrest BAI2 — classifies 2,318 bank lines and ties opening cash of $22,400,000 to the ledger, difference $0. The receipts forecaster re-fits payment timing for 24 named customers, the payments forecaster places 26 supplier runs and 6 payrolls, and the tax and debt scheduler places tax and debt service from signed sources only. Fourteen sources, 9,071 records, 6 min 48 s.
Receipts forecaster: “4 changed: Orchard +9 days, Sablewood +5, Corriedale +10, Yarrowdale +10.”
The checker re-adds 13 weeks × 8 drivers and finds Tidewell Foodservice's $1.2M counted twice in week 3 — sales order SO-88213 and invoice INV-604417 — and holds it. It asks a question about Sablewood Contractors: $0.53M in week 6 rests on its controller's email promising payment by 15 December, while its last 90 days average 11 days late. Marcus Okafor, the cash manager, settles the hold at 08:51, removing the duplicate, and answers the question: keep the promise and leave it in week 6, or follow the history and move it to week 7.
“The checker does not hold this number; it asks you to choose.”
Helen opens the forecast as a fan: the expected closing balance, its range and the $15.0M minimum. Week 6, w/c 14 December, closes at $14.2M, range $11.9M to $16.5M, and four weeks fall below the minimum at the low end. Revolver availability is $36.5M of $60.0M. Under each week, driver chips show receipts, suppliers, payroll, tax and debt service, and a Needs-you list holds the four things waiting for her.
“Closes $0.8M below the minimum. Two drivers: Tax −$6.1M, the federal estimated payment on Tue 15 Dec, and slower receipts.”
Week 6's −$6,100,000 is the US federal estimated tax, Q4 instalment, due Tue 15 December. The drawer shows the record, how it was built, the checker's verdict and its history, and quotes the source: Tax provision workbook FY26 v3, tab “Q4 estimates”, signed by Rachel Moreno, Tax Director, on Fri 6 November at 16:40. Confidence 0.99, checked.
The tax and debt scheduler never estimates tax itself: it uses the number Tax signed.
Orchard Retail, Net 45, carries $11.84M of open receivables. Remittance notes since 28 August say “paid in month-end run”, so its average over the last 90 days is 21 days late against 12 last year. Three invoices — INV-603881 for $1,240,000 due 4 Dec, INV-603902 and INV-603957 for $1,180,000 each — land in week 8, not week 6. One click drafts a request to Aisha Bello, the credit manager, to confirm Orchard's payment date by Fri 13 November.
“The agent drafts the request; the credit team decides whether and how to contact Orchard. No message goes to the customer from here.”
Last week's forecast for 6 November was $23.10M; actual cash was $22.40M, a gap of −$0.70M, −3.0% against a ±5% tolerance. Orchard paid $1.92M nine days late, freight carriers billed $0.83M over their orders, Tidewell paid $1.69M early for a 1% discount and a $0.38M state tax refund came early; $0.02M of bank timing is left unexplained. Three learnings wait for Marcus: Orchard from 12 to 21 days late, a 6.2% freight allowance, Tidewell 6 days early.
“The forecast never changes its own method. A learning applies from the next run once the cash manager or treasurer accepts it, and the decision is logged.”
Helen asks in plain words. The answer is drawn on the same fan in violet: if Orchard pays its three December invoices on time, week 6 closes at $17.8M instead of $14.2M, and the low point moves to week 9 at $16.3M. Then the caution, with its sources — Orchard's open invoices and its 14 remittances since 28 August.
“Orchard has paid 21 days late on average for 90 days, so I would not plan on it: the low end of week 6 would still be $15.5M, close to the line.”
She asks for a funding plan. FP-W46-1 draws $5.0M on Thu 10 December, before the tax payment, and repays $2.0M on Mon 11 January and $3.0M on Mon 1 February. With it the low point is $19.2M in week 6, the lowest P10 is $16.9M and availability never falls below $31.5M, far from the $7.5M springing covenant trigger. Estimated interest is $39.6K at 6.4% all-in, and the plan is within her $10.0M authority.
“Agents never draw or repay. The draw request goes to Halberd Commercial Bank only after you approve the forecast with this plan, and the cash manager releases it in the bank portal.”
Before she signs, Helen runs a prepared shock. If Orchard's receipts all arrive four weeks later, the low point falls to $10.0M in week 6, six weeks fall below the minimum on the expected line and nine at the low end, and the funding need is $8.0M. The $8.0M would be drawn on the revolver, with $36.5M available. She can save a scenario and compare up to two on the forecast.
Sign-off lists every check: opening cash tied, every week re-added, the hold settled by Marcus, sources current, every week above $15.0M with the plan, availability at least $31.5M. Helen chooses A2 · Approve with funding plan, writes a note to the record and signs. Version v3 is sealed and the pack goes to Daniel Reyes, CFO, to Priya Nandakumar in FP&A and to Halberd Commercial Bank in its 13-week format — more than a day before the lender's Tuesday 17:00 deadline.
“Approved by Helen Cho, Treasurer · A2 Approve with funding plan · Mon 9 Nov 09:42 · Version v3 sealed · pack delivered to 3 recipients.”
The same Monday forecast, seen by the five people who carry it — what their week looked like, and what it looks like now.
Seven agents forecast each driver, check each other and explain last week's gap; the treasurer decides and signs.
Reads prior-day statements from 4 banks and 11 accounts in BAI2, MT940 and camt.053, ties opening cash to the ledger cash accounts to the dollar and classifies last week's actuals by driver.
Forecasts customer receipts from open invoices, sales orders and how each customer really pays over the last 90 days, uses segment curves for smaller customers and flags any customer whose timing moved by more than 5 days.
Places approved invoices on the Tuesday and Thursday supplier runs, adds purchase-order commitments, payroll, capex, leases and intercompany, moves items off bank holidays and marks non-critical amounts as movable.
Places tax payments from the tax provision workbook Tax signed and the tax calendar, and interest, fees and amortisation from the credit agreements and their amendments, quoting the clause for each.
Re-adds every week and driver, looks for receipts counted twice, payments outside policy, stale feeds and customers whose timing drifted, and tests the expected and low-end balance against the minimum and availability against the covenant trigger.
Compares last week's approved forecast with actual cash every Monday, explains each gap by driver in one plain sentence with its evidence and proposes learnings.
Answers a what-if by changing only the drivers the question names, and when a week falls below the minimum drafts the smallest draw that keeps the low end above it, with repayment dates, its cost and its tests.
Approves and signs; CFO co-signs borrowing above $10M. The agents propose; a named person decides.
Ask about the forecast in plain words, or say what to test. Every answer cites its records, and nothing changes without your say-so.
What if Orchard Retail pays on time?
If Orchard pays its three December invoices on time, week 6 closes at $17.8M instead of $14.2M, $2.8M above the minimum. The low point moves to week 9 at $16.3M. But Orchard has paid 21 days late on average for 90 days, so I would not plan on it: the low end of week 6 would still be $15.5M, close to the line.
Why does week 6 dip below the minimum?
Week 6 closes at $14.2M, $0.8M under the $15.0M minimum, for two reasons: the $6.1M federal estimated payment for Q4, due Tue 15 Dec, from the tax provision workbook signed by Rachel Moreno; and slower receipts, because Orchard Retail's $3.6M due 4–11 Dec is forecast for week 8 now that Orchard pays at month-end, 21 days late. Payroll and supplier runs are normal for the week.
Draft a funding plan for week 6
Plan FP-W46-1: draw $5.0M on Thu 10 Dec, before the tax payment; repay $2.0M on Mon 11 Jan and $3.0M on Mon 1 Feb from January receipts. The low point becomes $19.2M in week 6, and even the low end stays above $15.0M every week. Availability never falls below $31.5M, far from the $7.5M covenant trigger. Estimated interest $39.6K at 6.4%. It is within your authority; I will not draw anything — the cash manager releases it after you approve.
Explain last week's variance for the CFO
Cash closed at $22.40M on 6 Nov, $0.70M (3.0%) below forecast. Orchard Retail paid $1.92M nine days late and freight surcharges added $0.83M; Tidewell paid $1.69M early for a discount and a $0.38M tax refund came early. $0.02M is unexplained — 97% of the gap is explained by driver.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds and rules every forecast is built and checked against are settings, and every change is logged.
| Setting | Default | Choose from |
|---|---|---|
| Minimum cash · Distribution · US | $15.0M | Set per perimeter, in USD thousands · treasury policy TP-04 |
| Variance tolerance · weekly closing | 5.0% | Any percentage; above it the variance explainer escalates |
| Variance tolerance · per driver | $1.0M | Any amount, in USD thousands |
| Confidence floor for automatic acceptance | 0.80 | Any value; below it a person confirms the number |
| CFO co-sign above | $10.0M of new borrowing | Any amount, in USD thousands |
| Receipts method, by customer | Behaviour: each customer's real payment timing | Behaviour (how they really pay) · Due date · Manual (cash manager) |
| Hold a number the checker cannot defend | On | On · Off — when on, a hold blocks sign-off until a person settles it |
| Apply learnings automatically | Off | On · Off — when off, a person accepts every learning |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “estimated” = our estimate (the split of the Monday hours; the $10M lower buffer is worth about $640K a year at the revolver's 6.4% all-in rate) · “typical” = published figures (AFP 2026 Treasury Benchmarking Survey — 49% name cash and liquidity forecasting their most challenging task; NeuGroup cash forecasting survey — most spend 8–16 hours producing the forecast, 93% compile it in Excel). People, companies, banks and customers named on this page are fictional — characters and sample data in the working solution.
Cash Flow Forecasting is cash flow forecasting software that builds a 13-week weekly forecast and a 12-month monthly view every Monday from your banks, ledger, payroll, tax and credit agreements. Seven agents forecast each driver, check each other, explain last week's gap and draft funding plans; the treasurer opens any number to its source and approves before the forecast goes to the CFO and the lender.
By the direct method, from the systems of record. The cash position reader ties opening cash to every bank account, the receipts forecaster places each open invoice on the day the customer is expected to pay, the payments forecaster places supplier runs, payroll and capex, and the tax and debt scheduler places tax and debt service from signed sources. Months 1–3 of the monthly view roll up from the 13 weeks; months 4–12 come from driver models on the FP&A plan.
The receipts forecaster fits each named customer's payment timing from its last 90 days of payments and remittance notes, rather than assuming the due date, and flags any customer whose timing moved by more than 5 days. When a late payer pushes a week below the minimum, it drafts a request to the credit team with the invoices, due dates and payment history. It never contacts the customer.
Every Monday the variance explainer compares last week's approved forecast with actual cash, splits the gap by driver and explains each part in one sentence with its evidence. Gaps below $0.25M with no cause found are left unexplained, never invented. It proposes learnings — a customer's new timing, a supplier surcharge — that apply only once a person accepts them.
Yes. Ask a what-if in plain words or run a prepared shock, such as the largest customer paying 30 days late, and the result is drawn on the same fan with its low point and funding need. The scenario and funding planner drafts the smallest draw that keeps the low end above the minimum, with repayment dates, estimated interest and tests against availability and the covenant trigger.
Yes. The agents build, check, explain and propose; they never draw, repay, release payments or contact customers. A held number blocks sign-off until a person settles it, every change by a person needs a reason, and the treasurer approves with a reason code, with the CFO co-signing new borrowing above the limit.
Yes. Minimum cash by perimeter, the weekly and per-driver variance tolerances, the confidence floor for automatic acceptance, the CFO co-sign limit, the receipts method by customer, holiday calendars, deadlines and recipients are all settings, and every change is logged.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your treasury policy, credit agreements, bank and ERP connections and a sample of past forecasts — and it goes live once every quality gate has passed. Before the first Monday run it back-tests on past weeks so you see its accuracy first; we will walk you through it on your own data.
Every covenant projected quarters ahead, so a step-down never becomes a default.
1-min demoHedge AccountingEvery hedge documented on the trade date and tested each quarter, so earnings stay steady.
1-min demoBank Fee AnalysisEvery bank fee checked against the agreed price, every overcharge claimed with proof.
1-min demoWe'll run Cash Flow Forecasting on your own bank files, ledger and last quarter's forecasts.