The ThoughtRiver Story: One Man’s Quest to Stay In The Zone

One Man’s Quest to Stay In The Zone

By legal AI contract analysis company, ThoughtRiver

The ThoughtRiver journey did not begin with a technology. It began with an idea: that it was possible to help overcome a growing problem faced by businesses everywhere – dealing with contracts.

As a lawyer and commercial director of a multinational company, Tim Pullan (pictured above), watched the problem escalate. If you look at a commercial contract from back in the 1980s, it will be a bit fuzzy and loose, more like a gentleman’s agreement,’ he says.

Contrast that with the ‘bloated’ and rigorous contracts that are the norm today. Size and complexity are driven by changes from the fashion for outsourcing to pressure from regulators. However, the language and structure of contracts has not kept pace with new types of business relationships.

‘It is like trying to construct a 100-storey building with 18th century techniques,’ Pullan says. ‘It has got completely out of hand.’ 

The burden of dealing with all these falls on the in-house legal department, where 70% of work is typically contract-related. Naturally, a general counsel will want to devote their team’s attention to the most important contracts, the big deals.  In reality they will be constantly distracted, with demands to check a stream of low-level contracts for potential impacts on the enterprise, its people or property. Unlike in a law firm, few in-house departments have junior lawyers on which such tasks can be delegated.

Once you’ve left the zone you’ve lost a large majority of the day doing things you don’t want to do,’ says Pullan. ‘It’s a huge opportunity cost.’

What is needed is a first line of defence – a screening mechanism to identify contracts that can be ignored without putting the business at risk.

Pullan’s blend of skills and personal experience led him to come up with a solution. He describes himself as a ‘child of the 1980s’, growing up with the first generation of home computers. His first experience as a developer came at law school, where he found himself working part time for a local practice which had installed a litigation automation application. I spent a few hours each week there – earning good money for a student – unearthing functionality they didn’t know was in the product,’ he says. 

On qualifying as a solicitor in 1998, he practised in technology law, specialising in data and privacy, before being recruited as commercial director at Experian Limited, a business at the leading edge of data analytics.

All this gave him a deep insight into the business of contracts, which he describes as ‘one of the last great vestiges of analogue information’.

‘I’ve negotiated contracts all over the world,’ he says. ‘One of the things that really slows it down is the medium, you’re trying to record precise information in an imprecise medium, that of language. It creates a huge amount of friction.’

The inspiration for ThoughtRiver arose out a conversation with a general counsel (GC). She was relating to me the familiar problem of being swamped, unable to do a proper job on the most important aspects. I said to her, ‘What if you had some automated system telling you which contracts are likely to be problematic: coding them red, amber and green, so you know the ones you need to look at and can ignore the rest?’ She almost fell off her chair,’ Pullan explains. 

Conversations with other GCs produced the same reaction.

By 2012, Pullan, then in Singapore with an international law firm, was working on a solution. It was a time of rapid advances in natural language processing and machine learning, the great enabling technologies of artificial intelligence (as distinguished from previous generation of rules-based expert systems). But it was not possible simply to go to a technology expert and say, ‘Build me this system’.

‘You have to remain incredibly close to how the technology is developed. I tried to outsource development at one stage,’ he recalls. ‘I wasted a bit of money trying to use various third-party products as part of the stack and it just didn’t work.’ The lesson here is that: ‘If you are looking to technology to produce deliverables that would otherwise be produced by a professional, to get it right you pretty well have to do it yourself.’

Tim Pullan (CEO), Emanuala Denaro (Legal Engineer), Rebekkah Provine (Lead CI Consultant), Elizabeth Sermol (CI consultant), Pete Nussey (Head of Business Development)

Another lesson learnt was that developing a product and serving a law firm’s clients are not comfortable bedfellows. Because the client’s demands have to come first, there is always the danger that the product development lifecycle will not get the care and attention it needs – and once that starts to slip the more work is needed to get the product back on track.

As a result, Pullan founded ThoughtRiver as a standalone business. ‘You have to be prepared to take a personal risk. That’s part of the journey,’ he says.

However, the law firm remained on board as an investor and one of its clients, a major advertising business, became an early tester of the ThoughtRiver system after seeing an early prototype.

In 2015, the business hired its first employee – developer Dom Hudson, now lead intelligence architect. The ThoughtRiver team is now 19, based in Cambridge and London, with clients all over the world.

Pullan modestly attributes success to a few lucky decisions. One was to go to the Cambridge-based Syndicate Room for the first seed-funding round. ‘It turned out to be the best decision we ever made. We got funding very quickly,’ he notes. 

Meanwhile, the business found itself leading a wave of interest in AI. ‘We found that if you put useful content out there, people came to us. If they come to you, you know they are genuinely interested, which made our sales process much more efficient!’ he says. 

But all this was underpinned by a simple, and highly attractive business proposition. ThoughtRiver’s core product is superficially simple: a system to structure and extract contract information and generate a single RAG (red-amber-green) score but underlying this is huge complexity.

Every contract review involves answering hundreds of legal questions based on hundreds of thousands of classifications and millions of items of training data. ‘This has nothing to do with robot lawyers,’ Pullan concludes, ‘We see our tool as being a little helper. Helping save lawyers time is what we do.’


 [Tim Pullan is the CEO and founder of ThoughtRiver, a pioneer legal AI contract analysis company. This is the first in a series of Sponsored Articles based on the ThoughtRiver team’s experiences working in-house and looking at topics including people, process and operating models.]