Legal prediction bot developers working on 80 per cent accuracy model

The Cambridge graduates behind LawBotX, the case outcome prediction bot combined with a lead generation tool for law firms, have taken a further step into the commercial world by rebranding their platform as Elexirr.

Relaunched last month as LawBotX, the bot is hosted on Facebook Messenger and also includes a machine-learning function. This refines predictions through use and filters prospective clients based on a law firm's usage history.

The bot's accuracy rate currently stands at 71 per cent. In a video posted on the business's website, managing director Ludwig Bull (pictured) said the team was now working on 'a next generation deep-learning model with which we hope to crack the 80 per cent accuracy'.

And so confident are Elexirr's founders about the system's performance that they have challenged lawyers to compete against it. The competition, which has its own #ElexirrLawyerChallenge hashtag, is scheduled to take place in September and will pit 'London's top lawyers in an epic man vs machine competition'.

A panel will give contestants real legal problems with a known outcome, such as a case that has already been decided, for consideration by both the human and the machine teams before coming up with a prediction as to outcomes.

The Elexirr challenge coincides with the launch of two new services on the back of the platform's reboot.

The first is referred to as a lawyer matching service, which will connect users to lawyers. Roll out is planned in September.

The second, LawBot Enterprise, is aimed at law firms and corporate counsels. One product, 'Legal Network', will allow firms to plug into the matching service. It will allow lawyers on the Elexirr database to accept 'matching' leads from the bot and to access the conversation the user has had with the bot about their claim.

The matching service will be based on preferences initially set by the firm. These can be based on location, budget, and practice area. The machine-learning component of the service will then kick in to only refer prospects based on the on the profiles of the ones previously accepted.

The widget will either sit on the firm's own website or the firm may accept referrals from users accessing the service from Facebook Messenger.

Another product, 'Corporate Analytics', will provide companies contemplating litigation with a similar prediction profile for their case. The bot's analysis will consider the characteristics of the case against data held 'in a large database of real legal problems with known outcomes' such as successful insurance claims.

Elexirr's next move will be to issue its own cryptocurrency via an initial coin offering (ICO) this autumn to fund its development. The blockchain-supported currency will be used by law firms interested in becoming part of the network.

Jean-Yves Gilg, editor-in-chief

jean-yves.gilg@solicitorsjournal.co.uk | @jeanyvesgilg