Premonition, the litigation analysis company, which launched in 2015, arguably has been at the forefront of what has now grown into a major segment of the legal AI market.
Artificial Lawyer recently caught up with co-founder, Guy Kurlandski (right in picture), to find out where Premonition had got to since its launch and where it was now headed.
The first thing to note is that Kurlandski is not your usual legal tech pioneer. He’s neither a natural language processing (NLP) coder, nor a 20-something law graduate.
But, Kurlandski is multi-talented. He’s a co-owner of the Lucky Player vodka brand, a joint principal in Lionheart Capital Management, a co-owner of several property companies, the CEO of a printing company, and of course, he’s now the CEO of Premonition.
When we Skype between London and Florida, where Premonition is based, Kurlandski reveals that he grew up in Primrose Hill, that quintessentially British part of North London with its pastel-painted Georgian homes. It seems far away from the massive condos and sunny beaches of Florida, as well as being a pioneer in legal AI.
We begin at the beginning. In 2008 Kurlandski moved to America and hooked up with an old friend Toby Unwin (left in picture), who had set up NetSearch back in 1998, a web-based headhunting platform. Unwin is now CIO of Premonition and equally colourful. Among his many accolades he is Honorary Consul in Orlando, Florida for the Republic of Austria. Yes, you read that right. He is also Chairman of the Municipal Guarantee Fund, which since 2007 has completed over $3bn in transactions.
Clearly, whatever was going to happen when these two non-lawyer entrepreneurs got together was going to be interesting. But, how did they come up with Premonition?
The story goes like this: Unwin had bought a strip mall in America. Soon after he was hit by a lawsuit by a claimant lawyer with regard to disabled access, which had been very much unexpected. The mall was also only single storey. The claim was fought and Unwin won, but was $8,000 out of pocket all the same.
The founders wondered whether the lawyer had a long track record of making such claims, but they had no way of knowing as the data needed to find this out was either hidden behind pay walls, or so badly organised there was no way they could analyse it.
Kurlandski notes, that yes, there are directories and the like, but as he puts it: ‘Is this guy really worth what we are paying him? Or is he rated because he won a big case 20 years ago?’ There was no way of knowing.
When they started to look at the big legal publishers they found a lot of data on case law, but there seemed to be no way to really drill down and get the actual information they wanted: was this lawyer any good, would he be the right person for this matter, would he probably win the case?
Moreover, when you looked at public records of cases the picture was a mess. There are 3,124 State courts in America. PACER, the public case data system, only has about 2% of all the cases.
To make things worse, if you actually could spend the time to approach multiple courts to find the bigger picture you found the data was completely disparate, with different courts recording case data in different ways and retaining different aspects of the outcomes.
Some systems didn’t even show who a defendant’s lawyers were. I.e. even though the judge in the court clearly knew this, the details were not recorded for posterity. In other cases, key data was replaced with a code number filing system. But, that filing system was only used by that court, in which case it was meaningless to anyone else.
Again, something had to change.
And so, Premonition was born, a system that allows users to see every available slice of data they could find on a lawyer, a court, a judge or anything else that could be scraped and analysed from court records. All that was needed was some good NLP, which Unwin was able to teach himself about. But, then they needed the data, which as seen, was a mess.
To cut a long story short, Premonition invested a lot of time and money and either gathered the data themselves or bought it from companies that had compiled it already, but had not thought to use NLP to create value from it. The company has grown rapidly and now has around 50 staff.
‘Now you can see who the best lawyer is versus X judge, on Y issue,’ Kurlandski explains. This can also show patterns of activity, e.g. if a lawyer or a client of theirs is making multiple or frivolous claims all over the country.
One fact they discovered from their research is that once a person sues in court some kind of invisible barrier is broken and they are then seven times more likely than others to come to court again.
Now, you might say, that’s very nice, but is that really helpful? Well, the answer has to be that if you have a choice of no data and some factual data that has been analysed objectively, which would you choose?
That said, some questions have been leveled at litigation analytics, such as: if many commercial claims are settled out of court and the details are often sealed, then how can we judge a lawyer fairly? And, if a client was expecting to lose, but the lawyer helped them lose less badly, then isn’t that a win?
Artificial Lawyer put these points to Premonition. Kurlandski replied: ‘It depends on the jurisdiction. 97% of US cases have a ‘Pre-Trial Outcome’, but there was still either a judgment or dismissal, so we assign winners based on outcome. In jurisdictions like the UK, we only have trial data, so that’s a valid criticism. However, who wants to hire the lawyer that often goes all the way to trial and usually loses?’
He continued: ‘Given the limited data available, aren’t you better off hiring someone you know is capable of taking a case to trial and winning? Surely they can negotiate a better settlement if that’s what you want? We don’t count a ‘good loss’ as a win. It’s a very slippery slope, though perhaps somewhat in keeping with the ‘everyone gets a ribbon’ American mentality.’
These points seem like fair questions, but with fair answers in response. Overall, the key selling points for any litigation analysis system are: why wouldn’t you want to know the win rate of a certain lawyer? Or the trend in a certain court toward a certain issue? Why not be better informed?
Whether in relation to litigation, or other areas where AI is focused, it seems likely the legal profession will steadily embrace better use of legal data in general. In which case, it would seem that the litigation analysis market will grow even if it is not a ‘cure all’.
Kurlandski adds that insurance companies that are continually being hit by claims are very interested in such data.
‘There are 15 million civil law suits filed every year in the US. That’s 41,000 claims a day and does not include small claims under $5,000,’ Kurlandski points out. ‘In England, the High Court [which hears large civil claims] sees just under 4,000 cases filed a year.’
This helps to explain two things: why US law firms focus so much on litigation and why America is the biggest legal market on the planet. It also suggests that anyone who can bring some insight into this mass of activity could be providing a valuable service.
This also explains why insurance costs in the US are huge compared to Europe, which Kurlandski estimates are about four times higher than in the UK.
So, what next? Premonition is a known brand in this segment, although there are now a growing group of competitors.
Kurlandski says they already have several international clients, though the UK has not produced many. This is perhaps because of the barrister system. Though, UK case data is also not that easy to work with.
The company is also now exploring how court cases impact company share values. I.e. a public filing that mentions a court case, but does not give exact details of the expected outcome in terms of costs or damages, is a hole in the data that Premonition might be able to fill with some predictive analysis.
‘We can also study data on auto crashes. What types of cars are involved, where do we see spikes? Court cases hold all kinds of data,’ Kurlandski explains. And where there is unstructured data and NLP there is valuable meaning to be found.
He adds that they could also build a database of all the cases that are never completed and settle before they advance beyond the local County Court. This would help to provide investors with a clearer picture of what may really be going on in a company or with regard to its products.
And, all this data can also be used by inhouse counsel to help inform their panel decisions for appointing external law firms. Again, the question is: this may not be the entire picture, but why wouldn’t you want at least some objective data on your advisers if you can get it?
This helps inhouse lawyers to go beyond the ‘IBM factor’ when choosing external counsel and make more fact-based choices.
To conclude, Kurlandski may not be a lawyer, or a traditional techie, but he appears to see very clearly the opportunities that using legal AI technologies can provide. Premonition is very new in terms of the time it’s been on the market and no doubt many other capabilities will be developed in the years ahead as the company explores the possibilities of legal AI.
For now it is part of a growing sector of litigation analysis/prediction systems working with AI technology. And, given the massive demand for litigation lawyers in America, then it would seem Premonition can predict for itself a bright future.