Much of the friction in the Lloyd’s and London insurance market comes from a familiar mismatch: business arrives in unstructured and inconsistent emails, documents and spreadsheets, while downstream systems require consistent, structured data.
Twiddling and tweaking data to push it through those systems has become a process in its own right, requiring considerable knowledge and expertise. Too much time is consumed making information fit rather than placing and servicing risk. Since that is unlikely to change any time soon, AI offers a useful and pragmatic way to bridge the gaps.
Modern AI tools are increasingly capable of extracting, classifying and normalising information from varied inputs, transforming it into usable data. The market is not short of standards and conventions. These are important because they provide the structured guidelines AI tools can use to process and transform data as it moves between systems.
Firms have been slow to embrace the opportunity
Some firms are already going down this route with some success. It's been a steep and sometimes bumpy learning curve, but the benefits have been tangible. Others have sought AI solutions through vendor platforms, but this approach can create a risk of platform lock-in, and the underlying AI may be opaque. Others are still grappling with assessing the risks and exploring all the options.
What appears to work well is an approach where AI is embedded in the workflow but loosely coupled to core platforms; operating in the gaps between the places where information is created and the systems in which it must ultimately be recorded. Its outputs remain visible, testable and subject to human oversight. This approach has the advantage of manageable incremental improvements, delivering benefits whilst more fundamental and strategic changes can be considered.
Placement and underwriting
At the placement and underwriting stage, the opportunity begins with broker submissions.
Underwriters typically have to work through emails and attachments, often presented in different formats and with varying levels of completeness. That takes time and creates inconsistency.
AI can extract key data items and map them to an agreed structure. It can also flag missing or inconsistent information before the submission reaches the underwriter. That means less back-and-forth, faster responses and greater confidence in the data supporting each quote.
Binding and documentation
At binding and documentation, the focus is on ensuring that what was agreed is reflected accurately in the contractual documents. This is not always straightforward. Differences can creep in between quotations, slips, endorsements and the final policy.
Before binding, AI can compare the relevant documents and flag discrepancies that could threaten contract certainty. After binding, it can check that endorsements and final policy documents accurately reflect the agreed terms. It can also support the production of policy documents using the same controlled, structured data.
The result is fewer errors, faster document production and a clearer audit trail between the agreed terms and the issued contract.
Delegated authority and bordereaux
In delegated authority business, the volume and variation of data become particularly apparent. Risk, premium and claims bordereaux arrive in different formats and often require extensive manual handling.
AI can extract the relevant information and map it into a consistent structure. It can also highlight anomalies, such as missing fields, unexpected values or inconsistencies, before they cause problems downstream.
Claims and settlement
Claims processes also involve large volumes of unstructured information, including notifications, adjuster reports, correspondence, invoices and policy documents.
AI can help classify and organise this material, extract important facts and financial information, identify missing documents and compare claim information with relevant policy terms. This can improve triage and give claims professionals faster access to the information they need.
Coverage decisions, reserving, settlement authority and payments should, of course, remain subject to appropriate professional judgement and controls.
Across all of this, the theme is consistent. AI works best when it quietly connects the dots between processes, improving the flow of data from one stage to the next.
That is what reduces delays, improves accuracy and helps the market operate more smoothly. It can also create the operational headroom and higher-quality data needed to support more fundamental market change in the future.
Wisereach helps organisations build the capabilities their people need to apply AI effectively. If you are interested in deploying practical AI solutions that challenge conventional approaches, get in touch.