Artificial Lawyer recently had a chat with Irish entreprenuer Gavin Sheridan to hear the unusual story of how start-up Vizlegal came to be and the potential the company has to disrupt the business of legal knowledge and research. (And how AI will play its part as well, naturally.)
Gavin Sheridan’s path to the world of legal start-ups is probably one of the more unusual stories. He’s not a disillusioned lawyer, nor has a PhD in machine learning. But, Vizlegal could have a major impact on the world of legal research in the years ahead.
Sheridan explains that he started off working for a media start-up called Storyful where he was the director of innovation from its foundation in 2010 until after its acquisition by News Corp in late 2013 for $25 million.
The company’s main role was to find and verify trending media content. For example this might involve finding a YouTube video that was soon to go viral, approach the person who filmed it and then arrange a deal with a media outlet, such as a newspaper.
Another example would be independently authenticating eyewitness videos from war zones like Iraq on behalf of other newsrooms. ‘News from noise’ was the company’s mantra, Sheridan explains.
He’s also been an avid blogger since 2002 and because of his writing and investigative work he has been faced with plenty of research problems. One area that Ireland-based Sheridan found it especially hard to get information from was via Freedom of Information (FOI) requests.
If you’ve ever tried to ask the government, in the UK or Ireland, for information under FOI then you’ll know how hard it can be. The usual response from government departments is to direct an FOI requester to a pile of documents which appear to help, but don’t really. Answers to specific requests for data can often go unanswered.
Nevertheless, Sheridan became very good at FOI requests as he learnt the best ways to carry them out through trial and error. And also dozens of appeals. Then one day something happened that would change the course of his career.
He was doing some research on NAMA, which is the publicly controlled Irish ‘bad bank’ that was formed after the financial crisis of 2008/9 to hold many of the bad debts from the Irish banking system.
The problem was that the Irish Government was rather touchy about NAMA and so excluded the body from the Irish FOI regime.
He had hit a brick wall, but Sheridan didn’t give up. He instead used the Irish equivalent of the Environmental Information Regulations (EIR), which comes from a European Directive and which had a different definition of ‘public body’ than that under the Irish FOI rules.
The case wound its way from the Irish Information Commissioner through two High Court decisions and ultimately a Supreme Court ruling in 2015, which found in his favour. NAMA spent five years fighting the Irish Information Commissioner and Sheridan over the definition of what ‘and includes’ means in Irish law, and lost.
Along the way he discovered something else that didn’t make sense: finding out even basic legal information about Irish or European judgments, case filings, or consolidated statutes, was extremely difficult. Either that or it could be pretty expensive.
While projects such as Bailii do provide the ability to search through large amounts of existing case law for free, Sheridan argues that simply publishing the text of judgments is no longer sufficient. The text, he says, must be enriched to become ‘data’, and tools must be built on top of that data to help legal professionals to do their work.
Information about litigation in Ireland is fragmented and hard to get hold of, or track. Even experienced lawyers are frustrated by the poor access to legal information. Moreover like in many jurisdictions, all that information – all those case filings, statutes, regulatory decisions and judgments – were not connected to each other.
That had to change, Sheridan thought. Here was all this useful information, but spread out and disconnected and with no easy way to make use of it.
For example, Sheridan points out: ‘Why are filings and judgments kept in separate systems, or at least not stored as structured and connected data, rather than text? Why are statutes so disconnected from judgments? Why do people end up with 100 tabs open on their browsers when researching relatively simple concepts?’
And even if you can find a case you are looking for its text is siloed in each document, rather than connected to any other meaningful information. And you may, of course, not know where to look in the first place as just using a keyword search will often not return what you’re seeking.
In short, there was no API – application programming interface – for legal knowledge, there was no intelligent search system, at least not in Ireland and few globally, that allowed easy and connected access to legal data.
What Happened Next
Sheridan gathered together a huge trove of Irish court decisions to produce a new integrated database which is designed to be easy to navigate, easy to read and could send simple alerts when phrases he wanted to track like ‘freedom of information’ appeared in new decisions, or when important FOI rulings might be cited by a judge, or when a particular judge issued a new ruling. That helped him keep track of the issues important to him.
The same was done for case filing data, for when journalists or lawyers wished to keep track of a particular litigation running through the system, or perhaps receive an alert when particular companies initiate or are subject to litigation.
Sheridan goes on to say: ‘The problem is often that official websites built by courts or parliaments see publication of text as their one and only job.’
‘They don’t make information available as data, and the user is often forgotten during the design of the [courts or government] website, leading to a smorgasbord of complex dropdown menus, tickboxes and search fields. And that leads to a poor user experience.”
The idea was thus to turn a non-responsive and inert mass of disparate legal text and information into something connected together and that could be made responsive and even proactive to a person’s needs, including suggesting new information that the user may be interested in.
Right now Vizlegal is in private Beta, with practitioners testing various features and feeding back on product development.
Sheridan plans to charge lawyers for access, but he says the prices will be extremely low in comparison to other systems, with pricing being transparent and cancellable at any time.
‘Pricing for most legal technology is opaque. This has to change,’ he also urges.
What Comes Next?
But, now that Vizlegal has started to accumulate all this legal data Sheridan sees so much more that could be done with it.
One particular area of interest is machine learning and AI, specifically around solving classification and recommendation problems. Sheridan has lots of plans in this area, but always with a focus on providing a practical use case.
The range of legal information in Vizlegal is also set to widen, including the addition of tens of millions of documents from multiple jurisdictions. For the UK, Sheridan is keen to expand coverage, and plans the addition of UK Supreme Court data in the coming weeks.
And if he is successful it will not just help lawyers, but could also help improve access to justice.
‘It’s not just a matter of copying text from one source to another. We want to enrich anything we get to allow for insight and analytics against every decision, and tools to help lawyers keep track of the ever expanding number of decisions. Anything that saves lawyers time and effort, while also delivering intelligence on the specific areas of interest to each practitioner is important to us,’ he says.
‘With Vizlegal we want to find nuggets in the disparate and fragmented sources of legal information globally. This is not too far removed from finding one important Tweet among millions as we did at Storyful, except this time we are seeking to create data where there currently is none. It’s a huge opportunity,’ he says.
Of course, this is just the beginning, but the ambitions are large. Thus far the company is self-funded, with the addition of being part of an accelerator programme in Dublin, Ireland, called NDRC.
‘We plan to be the largest data source for legal information in the world, so clearly we will need to raise more capital. But right now our primary focus is on our product and our customers,’ Sheridan states empathically.
And if Sheridan succeeds, especially once the AI capability is integrated into the system, then we really could be looking at something that could shake up the status quo of legal research and knowledge, which certainly seems like an area ripe for disruption.