We're five weeks into the first ever all-remote Lloyd's Lab - and a lot has happened so far.

The programme started on April 27th and lasts for a full ten weeks. The first half has been full of ice breakers, mentor meetings and workshops, and of course development of our proposed prototype.

As part of the Data and Models theme, a small team of our machine learning and product experts have been working with mentors to create an automated data extraction and enhancement email tool.

Our aim is to create an MVP that improves decision making and saves time for Marine Hull underwriters and brokers, bringing the speed in which the underwriter can respond to a broker's proposal down to a matter of minutes - rather than the hours or even days it can take currently.

Remote ice breaking

The Lloyd's Lab programme has previously been based in the Lloyd's building in London, where mentors and members can meet and network in person and get a feel for how the market works.

However, this year's pandemic made this impossible. The LMarks team, led by lab's Senior Programme Manager Lucy Coutts, has had the difficult task of making all sessions remote-friendly.

The first week of the programme was therefore mostly spent meeting our new mentors and fellow lab members on video conferencing tools.

We have been given the opportunity to work with some fantastic mentors: George Roberts from Sompo, Chris Lovick from MS Amlin, Adam Smith at Atrium, Ben Underwood from BMS, John Potter at Convex and several others.

Artificial's Head of Data and key member of the Lloyd's Lab team, Alexis Renaudin, said the access to mentors is one of the most valuable aspects of the programme:

'We are lucky to work with so many key individuals who can give us unprecedented access and insights to the London market - this would not have been possible without the Lloyd's Lab.'

During the initial ice breaker session, all new members and mentors had only 30 seconds each to introduce themselves to one another (think speed dating for insurtechs). Lucy and her team also organised Virtual Coffee Break and Friday drinks hours to keep up the informal feel of a face-to-face programme.

As well as these sessions, we learned more about the intricacies of the Lloyds Market and the unique way it operates, how risks are underwritten as well as a general introduction to the different ongoing initiatives about data and innovation and the future at Lloyds.

Refining our proposal

We spent the next week refining and articulating the proposal for our project. Each member must complete a project in the ten weeks, working together with our mentors to develop an interesting idea or concept into a prototype.

Our initial application to the lab was centred around automatic data extraction and augmentation from email attachments (PDFs, excels and other documents), so we put forward our ideas and spoke to our mentors about how best to execute the project.

In the first couple of weeks we spent time talking to them about their lines of business, discovering how we could bring value to them and how they could help us develop our tool.

After some deliberation, it was decided that our prototype should be focused on Marine Hull. Not only do most of our mentors have a connection to Marine Hull risks, but this line of business is what Lloyd's was originally founded on, so it felt like a fitting choice.

Developing the prototype

In week three we began to develop our prototype further. We know that Marine Hull underwriters need to access a lot of data to inform their decisions: this could be the ship's characteristics, where it travels to, when it was built and so on.

But it takes time to clarify with the underwriters exactly what information is needed to triage a risk effectively and replicate their logic.

To discover more, we held an informal workshop with one of our mentors from Sompo to talk through the information they would usually search for and input manually to assess a submission.

This really helped us understand the exact kind of data we would need for our tool and it allowed us to go out and find the right sources for augmentation.

The workshop also brought to light that many Marine Hull underwriters can spend 60-70% of their time going through submissions that don't even result in acceptance, giving us further vindication that our proposed tool is valuable to the market.

Processing sample data

As the weeks in the lab flew by, we began to process some information for our tool. With help from our mentors, the data team has been able to start testing the tool's data capture capabilities on sample emails.

We have been setting up the infrastructure for the tool and designs for adaptations to our existing front-end have begun.

The hunt for appropriate data sources has also begun to bear fruit, so we can start to augment the data from these sample emails using external APIs and develop the prototype further.

What's next?

We believe that our product is going to save valuable time and effort for underwriters and brokers. And it's not just relevant for Marine Hull; any complex commercial risk at Lloyd's can be similarly executed with the correct information and relevant data sources.

Faster appetite indication brings value to the market in different ways: with a layer of pre-filtering and automation, an underwriter can spend far more of their time saying 'yes' to high-value submissions and improve their risk portfolio.

Faster answers for brokers mean they can respond to customers quicker, giving them a competitive advantage in the market.

Ultimately, carriers who provide a faster, more accurate quote for their partners will be favoured in the market over those using old technology.

We are always on the lookout for more carriers and brokers to develop our product further. If you're interested in getting involved, please get in touch with Ollie Fox.