The current state of data

We've previously written about algorithmic underwriting and how it can help syndicates to use their data and make informed decisions. But the data available to insurers and brokers is rarely in the same structure or format.

Digital data standards

Two years ago we launched an algorithmic trading platform that is being used to write risks placed through the Lloyd’s market. In this case, the main enabling factor for implementation was digital API data from the broker.

Unfortunately, not all (or many, in fact) brokers are in a position to provide granular, risk-specific data points to underwriters digitally. The market continues to rely on emails with associated slip PDFs, schedules and other relevant information as the main transmission method during the underwriting or quoting process.

These documents take a lot of time for brokers to assemble and then for underwriters to subsequently disseminate and use as the basis of their decision making.

Data has to be structured in defined ways so that an algorithm or automation can be applied and used for decision making or support.

Until the brokers in the market are all fortunate enough to have a platform that allows them to create digital structured contracts, some level of manual intervention is required to translate the required decision making data points into the relevant carrier systems (for sanctions, rating, exposure, aggregation and other checks). 

Digital standards have been a topic in the market for many years, and we can now see the light at the end of the tunnel.

Adding value with machine learning

In order to drive efficiency with the current way of working we believe that using technology to extract data from submissions offers significant benefits that merit the investment from brokers and carriers alike.

Brokers can use the same technology to process the submissions they get from customers or producers, immediately adding value by creating the digital data sets for the carrier.

We have used our machine learning-based data extraction models to digitise and then process data found within submissions.

Whilst every document received will vary in structure, we have focussed our models on insurance contracts like MRC slip policies and schedules of value to ensure they’re able to handle what insurance professionals are accustomed to seeing.

This is then configured to process the extracted information based on the customers requirements. Once these data points have been collected (at the broker or underwriter stage of review) they are then available via API for re-use in any other application.

At this point underwriting clearance, sanctions checks, exposure and aggregation processing can all be executed and presented back to the underwriter to make more informed decisions or, if the relevant criteria is met, underwrite automatically.

The future of digital decision making

As the community gains more confidence in the technologies capabilities we hope to see a natural trend to more of the decision making happening automatically and the experts reserving their time and efforts for the risks that are the most complex in nature.

Our domain-specific language is used to model insurance products overlaid with the underwriter's appetite. The relevant information required for decision making is then extracted where possible and utilised to make decisions automatically or provide enhanced decision support to the user.

We believe that we’re the first company to create a specific language designed from the ground up to model insurance products.

This puts us in a unique position where we are able to support and enhance the existing underwriting submission and quote process whilst building a future-proof model that can be rapidly aligned to digital standards once the market reaches a consensus. In the interim we’re defining our own standards that are supporting the transition to unification.

If you would like to know more about our data extraction or algorithmic underwriting platform, get in touch with us.