InfoTrack, the conveyancing technology vendor claiming to support 25,000 UK conveyancers, has created a dedicated head of innovation and AI role in 2024. The position formalises work that began years earlier with AP1 form automation and has since expanded into what the vendor terms "document intelligence": cross-document validation tools designed to flag name mismatches, date conflicts and other transaction-blocking errors before they delay completions.
The operational question is sharper than the marketing. When AI extraction tools miss a title defect or flag a false positive that holds up a sale, who carries professional liability? InfoTrack's answer places responsibility squarely on the conveyancer. The vendor uses a "steering wheel" analogy: the tool gets you somewhere faster, but the user directs the outcome and accepts accountability for where it goes.
From extraction to validation
InfoTrack's early AI work centred on automating AP1 forms, initially using hand-coded rules before adopting machine learning models to parse documents more reliably. João Fernandes, promoted internally to the new head of innovation and AI role after seven years with the business, describes the progression as gradual rather than sudden. "We started doing this a long time ago," he told Today's Conveyancer. "People tend to think of AI as large language models, but long before that there was machine learning, decision trees, even OCR. We've been doing it for ages, we just hadn't named it as such."
The current iteration goes beyond simple data extraction. InfoTrack's platform now correlates information across multiple documents, checking whether names, dates and other key fields match. When it finds a discrepancy—say, a name spelled differently on a contract versus a mortgage offer—the system flags it early so the conveyancer can resolve the issue before exchange.
That capability has direct commercial value. Catching mismatches at the document review stage prevents queries from arising later, when they cost more time and client goodwill to resolve. But it also creates a new category of operational risk: what happens when the algorithm fails to flag an error, or when it generates false positives that slow transactions without cause?
Liability and the limits of vendor accountability
InfoTrack positions itself as "AI-enabled" rather than "AI-first," a distinction Fernandes says reflects the vendor's broader IP base, including council search data gathered by physical representatives and scheduling algorithms developed before machine learning became commercially viable. The framing also manages expectations about AI's role. "We don't present information as final or unable to be validated," Fernandes said. "The user has to see it, use it as a tool, and not just send it off without looking."
The platform includes verification features designed to make that review practical. Clicking the "source eyes" icon next to an extracted field shows the original document and highlights where the data came from. The intent is clear: give users the means to validate AI output without re-keying entire documents manually.
But the vendor's positioning—"aren't you responsible for the outcomes the technology produces?"—assumes a legal and regulatory consensus that may not exist. The SRA's competence guidance requires conveyancers to understand the technology they use and its limitations, but it does not define what level of AI transparency or accuracy disclosure is sufficient, nor does it clarify how PI insurers should assess risk when a firm relies on automated document checking.
InfoTrack's terms of service are not publicly detailed in the source material, so the contractual allocation of liability for AI errors remains unclear. If a name mismatch goes undetected and causes a completion delay or negligence claim, does the conveyancer's PI policy respond in full, or do insurers argue that reliance on unaudited AI output represents inadequate supervision? No case law or SRA disciplinary precedent yet answers that question.
What's missing from the commercial read
InfoTrack claims to support "over half the industry" based on its figure of 25,000 conveyancers. That implies roughly 50,000 conveyancers practising in the UK, though the basis for the count is unspecified. Individual users and firm licences are not the same metric, and monthly active users may differ substantially from total accounts. Without independent verification, the market share claim remains directional rather than definitive.
Equally absent: error rates, false positive and false negative statistics, or any benchmarking data for the document validation tools. Conveyancers evaluating the platform have no public basis to assess how often the AI misses a mismatch or how many flagged errors turn out to be benign variations rather than substantive problems. Vendors building competing tools face the same transparency gap, and none has yet made accuracy disclosure a point of competitive differentiation.
InfoTrack received the King's Award for Enterprise in Innovation in 2024, cited for modernising conveyancing through technology. The award does not specify whether AI tools were central to the assessment or whether the recognition covers the platform's broader feature set, including search ordering and matter management.