Aipedia: Building services engineers

AI in net zero and retrofit advice

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

AI can produce carbon and energy savings forecasts for retrofit and net zero plans quickly, but clients often read those figures as promises. Building services engineers giving this advice carry advisory exposure that can outlast the project by years.

Part of: AI in building services engineering

In short

Building services and energy engineers use AI to model retrofit options, forecast energy and carbon savings, estimate payback and draft net zero roadmaps for clients and portfolios. Forecasts depend on assumptions about occupancy, tariffs, grid carbon and how buildings are actually used. When savings fall short, clients may say they relied on the figures to invest. CIBSE Code clause 1.3 prohibits misleading statements, and clause 4.2 requires competence. PI usually responds to claims of negligent advice, subject to the wording, but guarantees of savings or carbon outcomes may go beyond what a policy covers.

How AI is used in net zero and retrofit advice

AI tools now analyse meter and BMS data, benchmark buildings, propose packages of fabric and services measures, forecast energy and carbon reductions, estimate capital costs and payback and write the client report. Portfolio clients may receive AI-generated roadmaps for dozens of buildings at once. Some outputs feed investment decisions, funding applications or public commitments by the client.

That makes the advice consequential. A forecast that looks precise to one decimal place can carry far more uncertainty than the report suggests.

The duties that apply

CIBSE Code clause 1.3: no statements known to be untrue or misleading. Presenting uncertain AI forecasts as firm outcomes risks breaching this. CIBSE Code.
CIBSE Code clause 4.2: only advise where competent or under competent supervision.
Statement of Ethical Principles: Principle 3 asks for the right expertise "especially in novel or emerging areas", and Principle 5 asks engineers to consider fully the risks of technologies such as AI. Statement of Ethical Principles.
Part 2A: where retrofit becomes building work, designer duties under regulations 11F, 11J and 11K apply to the design. Part 2A.

Why AI forecasts get treated as guarantees

AI-generated reports tend to present a single number for savings or carbon reduction, often without ranges or a clear list of assumptions. Clients then put that number into board papers, lender discussions or sustainability statements. When actual performance differs, perhaps because occupancy changed, the measures were installed differently or energy prices moved, the client looks back at the original figure. If the report did not explain its assumptions and uncertainty, the engineer may struggle to show the advice was reasonable when given.

Checking AI-generated retrofit recommendations

Review should go beyond the headline figure. Check that the baseline uses real metered data rather than benchmarks, that measures interact sensibly (fabric improvements reduce the savings available from heating upgrades, for example), that the electrical capacity exists for electrification, and that disruption, phasing and maintenance costs are included. Ask whether the tool double-counted savings across measures, which can happen in automated option appraisal. Where a recommendation depends on the building being used differently, such as reduced operating hours, say so explicitly and get the client to confirm it is realistic. A short, plain statement of the main uncertainties is often worth more to a client, and to your defence, than a longer report.

What can go wrong

Illustrative scenario: a consultancy uses an AI tool to produce a net zero plan for a client's office portfolio, forecasting energy savings from heat pumps, controls upgrades and lighting. The client invests on the strength of the forecast. Two years on, savings are well below the figure because the tool assumed occupancy and operating hours that did not reflect hybrid working. The client claims the shortfall and the cost of further works, alleging negligent advice.

How PI usually responds

A claim that advice was negligent is usually within PI, subject to the policy terms, but the loss claimed must flow from the negligence and the scope of the duty. Guarantees of savings, carbon outcomes or ratings, or energy performance contracts with financial penalties, may fall outside cover or be limited. Our guides to net zero advisory exposure for engineers and PI for energy consultants explain the exposure, and fitness for purpose versus reasonable skill and care covers why contractual promises matter.

Practical controls

Present forecasts as ranges with stated assumptions on occupancy, hours, tariffs and grid carbon.
Say clearly that forecasts are estimates, not guarantees, in the report and the appointment.
Check AI-proposed measures against site surveys and actual metered data.
Have a competent engineer review every client-facing forecast.
Avoid accepting savings guarantees or penalty terms without checking your PI.

How Apex helps

Apex arranges PI for building services engineers. A broker reviews your cover and how your policy treats AI-assisted net zero and retrofit advice, including any performance commitments.

Sources

Frequently asked

Are AI savings forecasts a guarantee?

Not unless you or your contract say so, but clients often treat them that way. Presenting forecasts as ranges with stated assumptions, and saying they are estimates, reduces the risk of a dispute later.

Which CIBSE duty applies to carbon forecasts?

Clause 1.3 prohibits statements known to be untrue or misleading, and clause 4.2 requires competence. Forecasts that overstate certainty or rely on unchecked AI output can engage both.

Does PI cover a claim that savings were not achieved?

PI usually responds if the claim alleges negligent advice, subject to the wording. If you guaranteed savings or accepted penalties, that contractual liability may not be fully covered.

Should I use AI for portfolio net zero roadmaps?

It can help analyse many buildings, but each recommendation still needs a competent engineer's review against real data. Errors in a portfolio roadmap can repeat across every building it covers.

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.