Hélder Santos first had a taste of tech in the legal world 25 years ago, when his brother asked if he could help his firm in Portugal with IT issues, such as combining Excel files. Today, he is Global Head of Legal Tech & Innovation at international law firm Bird & Bird.
While Bird & Bird is based in London, Santos works from Frankfurt, after a career path that has included CMS and KPMG Law in Germany.
Artificial Lawyer caught up with him to discuss all things legal tech and innovation. And what better way to start than the recently announced major POC with genAI-first startup Leya.
Onboarding GenAI
‘The team at Leya are young and very product focused. They have a commercial approach and want to be a partner. They also let us help them shape their product,’ says Santos.
And if you’ve not seen Leya before, one of its key capabilities is to tap law firm knowledge and leverage that with generative AI to help lawyers with a range of drafting needs. Hence, a POC across the firm is key, as lawyers across different practice groups need to figure out how it can help them for their specific needs.
‘We like the user interface and the way it gives citations. They also have explained their product road map and so we know what they will do, which is very different to others, which can be like a black box sometimes,’ he adds.
The company has also announced a partnership with FromCounsel, which provides in-depth legal content for commercial lawyers, with Bird & Bird taking part in their multi-firm pilot as well.
In short, with Leya they have found a genAI-first legal tech company they can really engage with and Santos clearly appreciates this. But, in terms of how this is helping the lawyers, how is Leya and the POC changing things?
‘This is extending the lawyers’ ability, and that also then brings more value to the clients,’ he says.
He adds that via the POC the firm is putting together a ‘library of shortcuts’ for how to best use the tool and lawyers are encouraged to share their experiences. They are also running workshops to expand product knowledge and give the lawyers the confidence to leverage it, with tech consultancy Simplexico helping out.
Before moving onto other areas, this site asked whether Leya is directly connected to the DMS – i.e. the Fort Knox of all the firm’s work product? Santos explains that the short answer is: no. The long answer is that lawyers can shuttle data from the DMS – wherever it’s allowed – into Leya.
He notes that eventually such a step may well happen, but that it will be ‘done in a structured and calm way’.
Overall, this is a big investment in time and lawyer attention. But, the view is that if you’re going to adopt a genAI tool across the business that can leverage so much knowledge and be fed into so many practice areas, then you need a full-scale POC, with training, workshops and knowledge sharing. There are no short-cuts here.
What Other GenAI Tools Are They Using?
Santos notes that like many other law firms they are making use of Microsoft CoPilot, although ‘it’s not perfect and there are a lot of gaps’ in what it delivers.
But, as they have it as part of their overall Microsoft suite of native tools, the view is they may as well keep it and find where it’s the right tool for the job.
In fact, Santos sees this stage of legal genAI as still very much about exploration and sometimes even waiting for access. They are planning on using Relativity’s genAI capabilities, they’re testing out new aspects of Luminance – which has moved from NLP / ML to now offering LLM outputs. And they’ve found that with translation tool DeepL they can do a lot more that language work, and are using it to also assist with drafting.
So, lots happening. But, AL has to ask: what about Harvey? Santos explains that in terms of what the firm wants they have ‘very similar use cases’ to Leya and so there is a degree of mutual exclusivity there.
Legal Tech Market Change
With so much focus now on genAI, what does he see elsewhere? Santos mentions that he’s well aware of how no-code tools have dropped off the radar and genAI has superseded some of that approach’s key use cases. In fact, while at KPMG he used to work a lot with no-code tools. So, he can see the change.
(That said, as Bryter told this site recently, there is still room for no-code tooling to be used in combination with LLM outputs, such as creating triggers based on a genAI tool’s response. And the Germany-based company is offering genAI features as well now.)
Santos is also intrigued to see what will happen with CLM as it brings in genAI and how that will mesh together with doc automation to integrate ‘data, process and standards’.
And he’s also interested in tools that provide diagramming, i.e. turning transaction information into visual data. (And that might be good news for companies such as StructureFlow. Plus, it’s worth saying that legal data visualisation is also going to be reshaped by genAI, via its ability to turn images into text.)
In short, even when an experienced legal tech expert such as Santos looks beyond genAI-first companies the impact of this new technology reappears again and again in areas where it didn’t previously figure. Moreover, all of this uptake of new tech has happened since the end of 2022. That’s an incredible shift within our market.
GenAI Standards
So, if genAI is going to dominate, then let’s talk about accuracy and the need for clear benchmarks and standards.
‘I am a huge fan of standards as they help people to measure [accuracy]. But right now it’s a bit messy,’ Santos observes and mentions the curious reality we have where Stanford University’s tests for legal research tool accuracy are seen as quite different from the tests some vendors use. Moreover, we have a situation where a company will say we have ‘96% accuracy’, for example, but then doesn’t explain what that ‘96%’ actually represents or how it was calculated.
Santos stresses that ultimately this is all about transparency.
And he’s totally right on that. We live in a world where benchmarks and standards are made public for dozens of products, from food to automobiles, but when it comes to genAI accuracy and legal work – which could cover some very high-risk matters – then we are still in what is mostly a grey area.
Transforming the Legal Sector
Moving onto change within the legal sector itself, how does he see things panning out? Will genAI be rocket fuel and lead to rapid change, after all it’s certainly changed the vendors?
‘Transformation [in the legal world]? It takes time. I have met a large law firm, not in the UK, which is still using File Explorer [which arrived with Windows in 1995],’ he notes, and mentions that there are some banks, such as in Portugal, that still work with the COBOL programming language.
In short, Santos is certainly not a believer in tumultuous, super-fast change in the legal world. Yet, as noted above, he is also passionate about embracing and experimenting with new technology – and doing so at scale across the firm.
Santos then highlights a key aspect of innovation in law firms in this context: ‘You only learn with mistakes, we need to fail to move onwards.’
I.e. here is a load of new transformative tech, and then here is an industry that is not designed to change fast, nor is naturally suited to experiments where things can go wrong as well as right.
And of course, in a tech environment ‘failing so you can learn’ makes total sense, in fact it’s inevitable, just as it is in scientific research. Or as a scientist would say: hypothesis, experiment, result, modify thesis, repeat and keep moving forwards.
Yet, in the legal world the goal is always to give the client an end product that won’t have any mistakes and where internal work patterns are generally predictable. One approach is about iteration and empirically-driven improvement that implicitly means ‘errors’ and the unexpected occurring along the way. The other is about a necessarily cautious culture that has embedded multiple layers of human error-spotting to produce a high-quality output the clients can rely upon.
So, how do you square these two patterns inside a law firm?
‘I think that sometimes we should reward failure,’ Santos states, that is to say, law firms should support experimentation, as every experiment, every test, every POC, allows the lawyers to learn and move forward.
Hence, if you are failing in a constructive way – and obviously not to the detriment of the clients – then you are (hopefully) getting better. And surely improving how the firm works should be rewarded?
Take for example SpaceX, every time it launches a new type of rocket the first few versions fail. But each one tells the company a huge amount and they celebrate even the launches that don’t reach space. They learn and then the next one is better.
Culture
Of course, this all needs a culture that supports a firm-wide approach to innovation that gives the time and energy to go through multiple learning curves, to see what works and what doesn’t, and then to keep going.
From this approach, Santos also notes, the lawyers learn a lot about any new tech and so they feel confident to talk to clients about it. Meanwhile, the clients want to know about areas such as genAI and what it can do. And, informed lawyers who really know what they’re talking about when it comes to tech, because they’ve been part of a detailed POC, in turn let the clients know this firm is on the ball when it comes to tech developments, such as genAI.
In short, it’s a ‘whole firm’ approach to innovation where the benefits extend outward to the clients.
To conclude, Santos says: ‘This is a journey and that takes time.’
So, whatever happens with the Leya POC and other projects the firm has running, Santos and team will learn from it and move forwards.
—
Artificial Lawyer is now on annual summer holiday until July 29. Have a good week and see you later this month.