In the future, will only some lawyers be able to beat the machines? Image source Universal Cable Productions
“When the economy does well, lawyers do well. When the economy does badly, lawyers do well” – Faculty of Law Professor, University of Cambridge c. 2001
I remember sitting in a crowded auditorium in the Law Faculty on West Road in Cambridge at the start my law degree. Three hundred law students sitting expectantly with their shiny new laptops, dictaphones, and notepads (the old fashioned, paper type) listening to these congratulatory remarks from the head of the faculty. Not only were we about to embark on the most prestigious legal curriculum in the world, but (and more importantly to us) we knew our future employability was guaranteed by one of the oldest and most highly respectable professions.
I was the black sheep of my class, being one of a handful not to go into practice as a solicitor or getting called to the bar. This was an active decision on my part – being much more interested in technology and the commercial world – I couldn’t imagine anything worse for me than to be stuck behind a desk, reading papers and debating nuances of small print my entire life. But at its height, the legal industry employed over 1.2m lawyers in the US. The number of practicing lawyers in the UK was around a tenth of this number, but many more people had been employed as paralegals and in other supporting roles. That was then, and this is now – and today many of my old classmates are finding themselves being made redundant, dazed and confused at the speed by which their jobs had been taken by replaced by machines.
* * *
One of the richest but least well-known women in the world, the rise to success for Lena Borysenko was swift and smooth. I had the good fortune to meet her at the beginning of her career when she was a new arrival to London’s tech start-up scene and when I was working as a mentor on an accelerator programme for tech start-ups with disruptive potential. Like me, Lena had chosen to avoid the conventional path of applying her University training to a professional life. A graduate from the technical faculty at the University of Kiev, Ukraine, she began studying how to apply deep learning methodologies to natural language processing, and in particular how to build automated response systems in order to provide massive workflow leverage. She was young, a bit geeky, and her command of English was fairly week. Although being a Natural Language Processing (or NLP) specialist, her knowledge of grammar and the structure of written prose was second to none. This being said, her pitch was poor, her business idea not very well articulated, and she had virtually a non-existent understanding of how it might achieve commercial success. Consequently, most investors and mentors avoided her, preferring instead to focus their attention on FinTech disrupters built on blockchain and other technologies rapidly ascending the hype curve.
Lena and I met a few times, and I recall spending most of those meetings trying to nudge her towards a commercial application of her product. Then, one day she broke all contact. “I be in touch in few months” read her text message. “I need go in the stealth mode – you see when I be ready”… And so Robottorney (www.robottorney.com) was born.
Now, seven years later in 2022 she commands a business with revenues in excess of $300m per annum, and unusually, being privately held, still retaining the ‘stealthy’ culture from which it was created. No-one has managed to put a valuation on Robottorney that is realistic, but were it to IPO, it would likely be one of the most valuable companies in the world – yet its reported they only employ around 1,000 people, mostly engineers, and mostly in Lena’s home country of Ukraine.
So, how did this all happen? Firstly, Lena realised that in order to find a market for her technology, she needed to focus on a niche. Secondly, she needed to find a niche which had been largely overlooked by technology providers. She realised the legal industry might well provide the human intensive work dynamic she needed, and so begun pitching to law firms to see how they could use her solution.
Now, fluent in English, she eloquently describes her early set-backs “They just weren’t interested in a ‘start-up’ solution with no existing customers” she said. “Everyone was reading the marketing from IBM around their Watson product, and this was the only thing getting budget”. Undeterred, she did something revolutionary and began her own law firm, initially offering cut price conveyancing. Supervised by a handful of experienced conveyancers, the bulk of the work being conducted by the machine. Launched in the UK in 2016, it quickly went viral and at a price point from £100 + disbursements, was a faster, cheaper, and more reliable solution than the competition – and profitable too.
Anyone who has ever bought or sold a property knows just how fraught an experience it can be. Robottorney simply took the worry away, and gave customers the confidence in its ability to de-risk the biggest investment of their lives. It was a clear winner!
One of they key aspects of the solution was that the machine learning algorithm could accurately predict whether the house purchase would fit into the set of scenarios that could be almost fully automated, and so was able to competitively price jobs that they could do. At the same time it could charge jobs that looked likely to be troublesome so high that customers went to the incumbent players. Every step of the way, of course, Robottorney built up a greater body of experience and became better at handling more matters, increasing revenue, profitability, and the all important word of mouth recommendation (it’s rumoured that Robottorney didn’t spend a single dollar on marketing in their first 3 years).
And so, Robottorney slowly grew, and by 2018 was cash-flow positive, Lena abandoning a Series-B round that was earlier envisaged. This was when the first inbound calls came from law firms (the sale team had been disbanded in late-2016 after the legal industry made it clear that they were not interested in buying anything without a big brand label). Lena remembers flying to New York for the first time to meet her prospect. “David Hardman was Managing Partner of one of New York’s most well respected boutique firms” she recalled; “Unlike his brother, Daniel, infamous from his public falling out with Jessica Pearson and Harvey Specter, he ran a clean shop, invested in a positive company culture, and in particular ensuring his team remained loyal to his firm”. While also not a big firm, Hardman LLC was an unlikely client for Lena, but intrigued, she made the trip anyway.
“It was clear from meeting him that he realised the industry needed to change”, she says, “while a fine lawyer, David is above all a first-rate businessman”. David’s idea was simple: Could Robottorney be adapted to work as a solution within a firm? Essentially, could it take the place of a trainee lawyer? “The answer was yes in spades” said Lena with a wry smile on her face; “This was the beginning of my dream becoming true”.
And so, David Hardman signed up his first six licences for Robottorney. While operating on the same core engine as the Robottorney conveyancing solution, a ‘Chinese Wall’ was created within the system so that data and ‘knowledge’ of matters could not seep into other parts of the system and thus avoid any potential conflict of interest. David struck a hard bargain. While paying $100,000 for each licence and signing up for three years, but in return during the initial term Robottorney agreed to not sell the product to any other US firm (the deal did not include UK firms, much to Lena’s good fortune).
The economics were simple. While more expensive than hiring a trainee lawyer out of law school, the Robottorney licence took the place of the trainee. Four other trainees were repurposed to overseeing the work done by Robottorney, helping the machine learn from its mistakes and becoming every day a better lawyer. “I had the equivalent of a team of 10 that was costing me more than if they were all human lawyers”, explained David; “but working together, within twelve months, they were achieving the same throughput as a team of 15. By the second year, we were getting a 2:1 time leverage”. This represented big savings for Hardman LLC, which he put to use investing in adding a couple of rainmakers to his senior team to help win more business. “What swayed me wasn’t the efficiencies that an AI could provide, it was the avoidance of hassle”. Having witnessed the chaos within his brother’s firm caused by reputational damage of unqualified staff, and periods of poor morale, etc – David realised the machine would never fail him. “It just plugged away, day and night. Never took a day off sick, never caused an HR complaint, and always remained focussed on the task” he said; “When it made mistakes, these were easily picked up by the partners in the firm. We simply developed a model of oversight to ensure all the work conducted by Robottorney was reviewed by senior staff before it went out”.
The human lawyers quickly shifted from lawyering, to review;
From delivery, to sales;
And poignantly, from profit centres, to cost centres…
With Hardman LLC as the poster-child client, Lena was able to quickly win business in the UK market. In the last two years, 40% of the UK Top 100 law firms have become clients. Robottorney still provides direct to consumer services for conveyancing, but this receives much less focus from top management these days. “Dominating the supply of services to the legal industry has always been our top priority”, explains Lena. “We’re going to focus on this, and then adapt the solution to other markets”. Its estimated that by the end of the decade, over 50% of ‘white-collar’ jobs could be replaced by systems such as Robottorney.
With Robottorney now selling licences for just $10,000 per seat, it has decimated the legal profession. Enrolments to Law Schools have dropped by 80% and schools have become even more highly selective on the places they give (I would never have matriculated under the new rules, that’s clear)!
With 30,000 virtual ‘lawyers’ being sold by Robottorney, it’s become the largest supplier of legal services in the world, but only has 100 lawyers on its own books. Reinvesting its profits into sales and engineering, Robottorney plan for high top-line growth over the next five years compounding the problem further for the UK economy. For with so many human lawyers out of work, the UK Treasury has seen a sharp decline in income taxation revenues. Not only were there lots of lawyers just a few years ago, but they were also highly paid – therefore in the highest taxation brackets. Now, the average salary for lawyers has dropped by 50% with so many unemployed fighting for jobs. Less lawyers, lower salaries, lower tax revenue for the state. Factor in the potential for higher graduate unemployment, and the Treasury is faced with a significant problem.
Still, Lena is doing well – and so is David Hardman. Being a first mover into Legal AI has enabled his firm to rocket up the US Top 100, and his overheads are a fraction of his competitors. If it wasn’t for his rapid growth trajectory, his firm would easily be the most profitable in the industry.
It amazes me how ill prepared we are for the economic impact of AI. Most articles on the subject crassly ask “will Robots steal my job”? It’s not robots we should be worried about, its the people who own or license the IP of the AIs which can be scaled across entire industries in a few years, thereby concentrating wealth in a handful of investors at the expense of the wider economy. The question we need to be asking ourselves is “how can I share in the new AI economy”? Paranoia about own employability is likely to be an apocryphal endeavour, focussing on how to share in the wealth distribution and benefit from the greater freedoms of a shorter working week – now, surely, that’s an opportunity for everybody – not a threat?
That’s why I believe that AI start-ups ought to be encouraged to seek crowd-funding to get themselves launched, and state-sponsored funds for AI should also be ramped up. Unless this happens governments are likely to have a greater mess on their hands than any Hollywood malevolent-AI movie can conjure up – and the rest of us will be reminiscing, not reproaching our 9 to 5 office jobs.
I’m keen to explore this topic further and hear your thoughts too, even if you do think AI is still a bit all too science fiction?
Let’s start the debate… what do you think?