Aipedia: Management consultants

AI implementation projects that fail to deliver

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

When an AI adoption programme led by a consultant fails to deliver the promised benefits, the client may claim negligence, breach of contract or both. How the statement of work described AI outcomes often decides which, and whether PI responds.

Part of: AI in Management consultants

In short

Some consultancies now lead AI adoption programmes: selecting tools, redesigning processes, training staff and tracking benefits. These projects are prone to scope creep, shifting technology and benefits that depend on the client's own data and people. If the programme disappoints, the client may say the consultant was negligent, or simply that it did not deliver what was promised. PI usually responds to negligence, subject to the wording, but commonly excludes guarantees, performance warranties and pure contractual failure. Careful SOW drafting on AI outcomes protects both the relationship and the insurance position.

What AI implementation work involves

AI programmes typically start with an opportunity assessment and business case, then move to tool selection, pilots, process redesign, change management and scaling. Consultants often lead the programme office, write the benefits case, manage vendors and report progress to the board. Some also configure tools or build prompts and workflows themselves.

The benefits are often expressed as hours saved, headcount avoided or faster turnaround. They depend on data quality, user adoption and vendor performance, much of which is outside the consultant's control.

That creates a mismatch. Clients buying an AI programme often want certainty and may press for fixed outcomes or payment linked to savings. Consultants keen to win the work may agree. The terms negotiated at the start then decide, years later, whether a disappointing result is a professional negligence question or a pure contract dispute.

Why AI projects slip

The client's data turns out to be incomplete or inconsistent, so tools underperform.
Scope grows as stakeholders add use cases, without formal change control.
Vendor products change mid-project or do not perform as demonstrated.
Staff adoption is low, so time savings stay theoretical.
Legal or data protection issues, such as automated decision rules, emerge late and force redesign.

Our guide to scope creep and PI claims covers the general pattern.

Duties and contract terms

The ICMCI AI Code requires meaningful human review (3.1), transparency (3.3) and engagement terms that cover AI scope, IP, data protection and restrictions (7.3); see the Code text. Where the programme involves automated decisions about people, the Article 22A to 22D regime in force since 5 February 2026 applies to the client's design (Article 22C).

The contract matters most. A SOW that commits the consultant to "deliver" a percentage saving or "ensure" a tool works reads like a guarantee. One that commits to "advise on", "support" or "use reasonable skill and care to help achieve" sets a negligence standard. Dependencies on the client and vendors should be written down.

What can go wrong

Illustrative: a consultancy is engaged to roll out generative AI tools across a client's customer service function. The SOW states that the programme will deliver a fixed reduction in handling time within a year. The tools work in pilot but adoption stalls, data issues limit accuracy and the target is missed. The client withholds fees and claims damages for the savings it says it was promised, alleging both breach of contract and negligent programme management.

The consultant may have managed the programme competently. If so, the negligence claim may fail while the contractual claim on the target succeeds, and that is where insurance questions arise.

Negligence versus non-performance in PI

Consultants' PI usually covers liability for breach of professional duty, including negligent programme management, subject to the terms. Some wordings may exclude or limit:

guarantees, warranties and liquidated damages beyond what the law would impose without the contract (see contractual liability in PI);
failure to meet a timetable or performance target where there is no negligence;
return of fees and the insured's own costs of re-work.

So a project that fails without negligence may leave the consultant uninsured for a contractual promise. The distinction between strategic advice and implementation also affects how claims are framed. See AI contract clauses and indemnities.

Practical controls

Describe AI outcomes as targets the client owns, not results you guarantee.
List client and vendor dependencies in the SOW.
Use formal change control for new use cases.
Report risks to benefits early and in writing.
Cap liability and exclude consequential loss where you can.
Show any outcome-based terms to your broker before signing.

How Apex helps

Apex arranges PI for management consultants and can review how your wording treats contractual promises on AI programmes. See the AI in management consulting overview.

Sources

Frequently asked

Does PI cover an AI project that simply missed its targets?

Usually only if the failure was caused by negligence. PI wordings may exclude guarantees and failure to meet performance targets without negligence. If your contract promised a specific outcome, that promise may be uninsured, depending on the wording.

How should a SOW describe AI benefits?

As targets the client owns, supported by your advice and reasonable skill and care, with listed dependencies on the client's data, people and vendors. Avoid words like guarantee, ensure or deliver for outcomes outside your control.

Who is responsible if a vendor's AI tool underperforms?

Primarily the vendor under its contract with the client. A consultant may face a claim if it negligently recommended the tool or failed to test it properly. Clear tool selection records and written caveats help.

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.