In another in our series of articles reflecting presentations at the SCL Tech Law Futures Conference, Karl Chapman describes an approach that will enable you to keep your options open so that you can ride the AI wave
It is widely claimed that, when giving directions, cabbies suck their teeth and say 'I wouldn't start from here …'. We're often struck, inevitably unfairly on occasions, by a similar thought when we either read about the current and potential use of Artificial Intelligence (AI) in the legal market or discuss developments with potential customers or solutions providers.
We are believers
Putting to one side the challenge that AI means many different things to many different people, I should state right up front that we are believers. We have absolutely no doubt that AI and related technology trends will have a significant impact on all markets and value chains. The legal market, globally, is not immune from this.
As a result, the Riverview Law Board has placed its bets and organised Riverview Law accordingly. We have invested millions of pounds in Kim, our New Jersey based knowledge automation and workflow platform.
Kim is a virtual assistant to knowledge workers in all sectors. Through its no-code configuration model, Kim enables knowledge workers, including lawyers, to take control of their critical business processes and work to make better and quicker decisions. Kim does this by utilising its patent pending artificial intelligence, assimilation and, critically, data management capabilities. We believe that the direction of travel is clear, even if timing is unpredictable, and have acted accordingly.
It is from this position of belief that we have what is hopefully a helpful reflection. From much of the activity that we see in the legal market we suspect that, with exceptions, many organisations are starting their AI journey from the wrong place (the outside in). We sense that much energy, money and time is being (and will be) wasted chasing fashion-driven AI solutions rather than proper investment, problem solving and experimentation. Indeed, much of the effort we see in this area can be put in the 'being seen to be doing something' box. At the end of such a process, the inevitable happens. Expectations are disappointed. The technology is blamed. Future investments are delayed or shelved.
This chain of events is potentially life-threatening to organisations. Failure to understand and ride the current data, IT and AI wave, over time, risks all. It is a 'bet the firm' issue … particularly if the wrong bets or no bets are placed!
Given the technological and data revolution we are living through, inaction is highly dangerous. But the irony in legal is that the risks can be reduced and the opportunities of success increased by learning from other industries, and in particular from house builders.
We can learn a lot from house builders
Typically you would not build a house from the roof down. But that is what we see happening in many law firms and organisations, with some notable exceptions (Vodafone, Barclays, Allen & Overy, DWF, PwC). There is a temptation to seek instant solutions. Under pressure to respond to noise and market trends, companies appear to be trying to put the AI roof on the house without building or re-constructing the foundations, walls and floors first. As one leading market commentator observed when I discussed this with him:
'You're wrong. They aren't even building the roof. Many have just bought some
tiles and are now trying to work out what to do with them.'
By the way, we recognise this process because this is exactly what we tried to do when we started our AI journey in 2013. We made many mistakes and learnt many interesting lessons. Which is why the house building analogy resonates so strongly with us now.
If you have ever watched a house being built from scratch, two things will probably have struck you. Firstly, a long time seems to pass with very little progress being made even though there is a full building crew on-site. Then, suddenly, when the foundations are finished, the house takes shape remarkably quickly. One morning the site was just concrete foundations and ground works and the next day the house starts to emerge, proudly, from its rock-solid foundations. It is almost as if it is saying 'I was always here but you couldn't see me'.
Secondly, in those weeks when you saw little progress being made, you were probably also surprised by how small the building footprint was. Clearly, a relatively small house is being built on the plot. But, when the house starts going up you are amazed by how big it actually is given how small the foundations and footprint appeared.
This is where the fundamental role played by the foundations comes into play. In the house that has been built there are many different rooms; lounge, dining room, kitchen, bedrooms, bathroom … The four bedrooms are all different; in shape, decoration and furniture. But, critically, all the rooms, with their different functions and designs, are built on the same foundations.
Enabling a sustainable legal AI solution
Think of this analogy in the context of building a sustainable legal (or indeed any) AI solution. Before building the AI roof, before even buying the tiles, it is important to put solid and durable foundations in place; (i) the data-layer, (ii) workflow and process automation and (iii) reporting and dashboards. Combined we call this the 'Foundational Data and Context Layer'. Using the instruction and triage process as an example, the legal data layer will include data points that show:
· • who the instruction came from;
· • what business unit they are in;
· • what the work type is;
· • what the urgency of the work is;
· • whether the instruction is complete;
· • when we received the instruction;
· • who we allocated it to;
· • what its current status is;
· • how long each stage of the work took … etc.
This is the foundational data layer upon which the workflow process and the real time and trend data is built.
Continuing the analogy, with these triage foundations in place, which are relevant to all legal work types, it is relatively easy to build the equivalent of the walls, floors and rooms in a house. Rather than the lounge, dining room, kitchen, bedrooms and bathrooms we have contract management, litigation, employment, property, IP etc - different work types built on the same foundational data layer with their own context (their subject matter data layer) on top.
With the foundations, walls, floors and rooms built, it is now easier to both enable existing third-party AI solutions and build tailored AI solutions using tools like Kim. To build context from the inside out, not the outside in. To create context from the foundational layer up, rather than by trying to ingest a huge corpus and then infer context. The latter approach works well with a global corpus such as diseases. It struggles when context differs by jurisdiction, geography, work type and organisation. For example, banks in the UK are subject to the same laws and regulations but even with this common environment each bank has a different language (dictionary), risk appetite, business model, data layer. Context is organisationally specific.
The hard yards
Building this 'Foundational Data and Context Layer' is not easy. The foundations represent the very hard but invaluable yards that need to be navigated.
Like that new house build we have spent a long time creating our foundations. We now have the Riverview Law 'Foundational Data and Context Layer' built in Kim. It enables us to execute at a pace that we never thought possible. A bit like those comedians who appear on television and become overnight successes after they have spent ten years learning their trade on the club circuit, we have now done our club tour and can go global with Kim.
We have made many mistakes along the way, particularly in 2013 when we started from the outside in and tried to build the roof! But the last year has vindicated all the emotional energy, money and time invested.
Place your bets
One of my friends, who is a technology expert, recently said to me - 'There is no AI magic wand … yet'. The emphasis on 'yet' is important. The direction of travel is clear.
So, create your 'Foundational Data and Context Layer' and keep your options open so that you can respond to whatever happens next. Place your bets.