Insurers Are Still Piloting. The Defendant Side Just Industrialised.

Two numbers landed in the same week. Read together, they say which half of the credit hire file is moving faster.

What actually launched this week?

On 7 September 2026, QuestGates launched an integrated counter fraud business, QuestGates Counter Fraud Solutions, bringing data analytics, investigation, forensic accountancy, surveillance and legal work under one service. The firm names credit hire fraud operators specifically among its more than 175 specialists, and says it already acts for 18 of the top 20 UK personal lines motor insurers. That is one week after QGLaw, QuestGates' legal arm, hired Gary Herring from Keoghs as Partner and Head of Credit Hire. Two moves in eight days, from the same group, on the same part of the file.

We are quoting QuestGates' own figures here, including the claim that it delivered more than £40m of fraud savings for large property insurers in 2025. Those are the firm's numbers, not independently audited ones, and we say so because a claims operation reading this should treat them the same way: as a resourcing signal, not a verified fact.

How does that compare with where AI adoption actually is?

Slower than the headlines suggest. Accenture's research across 263 senior insurance executives, commissioned research rather than an independent study, found that only 23% of insurers have achieved what it calls enterprise-wide AI integration. That is despite 81% reporting a measurable improvement in gross written premium, at least 5%, from AI and data initiatives already in place. The barriers named most often were legacy system integration (50%) and data quality (45%). Seventy per cent run a targeted AI skills programme. Only 14% have scaled one across the business.

So the value is proven and the rollout is stuck. That is not a contradiction. It is what happens when a large organisation tries to run one AI programme across every department at once, rather than solving one well-defined problem at a time.

Why does this matter specifically for a credit hire file?

Because the two numbers describe opposite ends of the same dispute. QuestGates and QGLaw are resourcing up the challenge, on rates, on periods, on need for hire, with named specialists and a stated fraud focus. Most claims operations, on the evidence above, are still running pilots rather than production tools on the response side.

That is the gap a narrow, well-scoped tool closes faster than an enterprise programme ever will. A system built to do one thing well, verify a rate, structure an evidence file, flag a gap before it becomes a dispute, does not carry the legacy integration and data quality problems Accenture's respondents named as their top two blockers. It was built for the job, not adapted to it. That is the whole design brief behind CreditHire Assist.

What should a claims operation actually do with this?

Three things, none of which require an enterprise programme.

First, treat the file at first notification as the moment that matters, not an afterthought. A better-resourced challenge on the other side means the evidence gaps that used to go unnoticed will not stay unnoticed.

Second, keep the audit trail. If a rate or a period gets challenged, the operation that can show its working, dated, sourced, consistent, is the one that holds its ground. That is a workflow discipline, not a model capability. It is also why we keep arguing that the number on the dashboard has to be able to say no.

Third, resist the instinct to wait for a bigger AI rollout before fixing the parts that are actually costing money now. Accenture's own data says the constraint is plumbing and data quality, not the technology itself. A scoped tool sidesteps that constraint rather than trying to solve it.

Is there anything else moving on the process side?

One item is still unconfirmed and worth naming plainly. GTA ADR Phase 2, binding resolution for credit hire invoices under £10,000 unresolved 61 days after presentation, was due to begin this month. As at the second week of September, no trade source or industry publication has confirmed it has actually gone live. We are not treating it as active until it is, and neither should anyone reading this.

If you want to talk about where a scoped tool would help your operation before a wider AI programme gets there, get in touch.