Aipedia: Accountants

AI cash flow forecasts and business plans

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

Accountants can use predictive AI to build client cash flow forecasts, but they must explain the limitations of the work under R113.3. When a forecast supports borrowing or a decision to keep trading, a flawed model can lead to large reliance claims.

Part of: AI in Accountants

In short

AI forecasting tools project cash flow from historical ledger data, seasonality and pipeline information, and can produce a polished business plan for a funding bid in hours. The ethics Code requires accountants to understand the technology, to recognise that the data may not be sufficient for it (300.6 A2), to decide whether its output is appropriate for the purpose (R320.11) and to explain the limitations of the work (R113.3). If a client borrows or keeps trading on an over-optimistic forecast, the loss can be large, and PI cover usually depends on how the engagement and reliance were framed.

How accountants use AI for forecasting

Forecasting tools linked to cloud ledgers can project receipts and payments from past patterns, flag likely cash pinch points and run scenarios. Firms use them for 13-week cash flows, funding applications, business plans for investors and going concern support for directors. Generative AI may then write the narrative that explains the assumptions.

The model is only as good as the data and the assumptions. Historical patterns do not predict a lost customer, a price rise or a change in payment terms, and a machine-learning model may not show how it reached its numbers.

The duties that apply

R113.3 in the IESBA technology revisions: be aware of the limitations of the work and explain what they mean. For a forecast, that means stating the key assumptions and what the model cannot predict.
300.6 A2: the data available might not be sufficient for effective use of the technology, for example a short trading history or poor-quality ledgers.
R320.11: decide whether the output is appropriate for the intended purpose, such as a bank application.
113.1 A3: competence includes understanding the technology you use.

ICAEW enforces these provisions from 1 July 2025, as announced in April 2025.

What can go wrong

Illustrative scenario: a practice uses an AI tool to prepare a 12-month cash flow for a client seeking a loan. The model extrapolates a strong previous year, but the client's largest contract is ending. Nobody overrides the assumption, and the forecast pack does not say the model relies on historical patterns. The loan is granted, trading falls short, and the directors later say they would not have borrowed or continued trading on a realistic forecast.

Forecast claims are hard because the accountant did not promise the outcome. The argument usually turns on whether the assumptions were reasonable, whether they were explained, and who was entitled to rely on them.

Going concern support is a particular pressure point. If directors use an AI-built forecast to decide that the company can keep trading, and it later fails, an insolvency practitioner may look at the forecast and at the advice around it. The firm will want to show that the assumptions were tested against what the directors knew, that the model's limits were explained, and that the decision to trade on was the directors' decision, not the tool's.

How PI usually responds

A claim that a forecast was negligently prepared generally falls within an accountant's civil liability PI, subject to the terms and excess. Where a lender or investor relied on the forecast, the first question is whether a duty was owed; see Caparo for accountants. Engagement letters that limit reliance and set out responsibility for assumptions help; see engagement letters and liability caps. Forecasts for large borrowing can produce claims that test limits, so review how much PI cover you need.

Practical controls

Agree the key assumptions with the directors in writing and record that they are the directors' assumptions.
State the model's limitations on the face of the forecast.
Override historical trends where you know of a change, and note why.
Keep the model version and the inputs used.
Apply the same care to management accounts commentary and to any valuation that uses the forecast.

How Apex helps

Apex arranges PI for accountants who prepare forecasts and business plans, and a broker can check limits and wording against this work. See the accountancy AI overview and whether PI covers AI mistakes.

Sources

Frequently asked

Do I have to tell the client a forecast was built with AI?

The Code requires you to explain the limitations of the work, which in practice includes how the forecast was built and what it cannot predict. The PCRT guidance and the CCAB draft also favour telling clients how AI is used.

Who owns the assumptions in an AI-generated forecast?

Usually the directors, but only if the engagement letter and the forecast say so and the directors have agreed them. If the accountant lets the tool set assumptions without discussion, the client may argue the accountant took responsibility for them.

How should we test an AI forecast before issuing it?

Compare the projection with known events such as contract renewals, price changes and payment terms, run downside scenarios, and reconcile the opening position to the ledger. Record what you changed and why, so the reasoning behind the final forecast is visible later.

Will PI cover a claim from a lender who relied on our forecast?

It may, if the firm owed the lender a duty and the policy covers the service, subject to its terms. Many claims fail at the duty stage. Clear reliance restrictions in the engagement letter and on the forecast itself reduce the risk.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.