There is beauty in simplicity and repetition
by Will Darbyshire
LEVERTON is a company, founded in Germany, that uses artificial intelligence to help us better understand the information found within complex, unstructured documentation. Their clients in the UK market already include JLL, Savills and GVA, all of whom rely on LEVERTON to more efficiently, and more affordably, understand the data sat within their legal papers, contracts and lease agreements. I met with Richard Belgrave, LEVERTON’s Head of Europe, to find out how, through the complexities of machine learning, they are providing an uniquely focused service which, although difficult to conceive, is beautifully simple to understand.
Simplicity is deceptively complex. As Steve Jobs once said, “you have to work hard to get your thinking clean to make it simple…once you get there, you can move mountains.”
In recent years, this is a philosophy that PropTech has failed to adhere to. I lose count of the number of companies I’ve spoken to whose USPs are either so complex or so tenuous that their value is hard to find.
So common has this become that, when I was first researching LEVERTON for this article, I was hoping there must be more to it than I could see because, as I write, I wasn’t really sure there was much to be said. But that was only until I realised that it’s LEVERTON’s simplicity, one which was, and still is, incredibly complex to achieve, that is driving their already enviable success in the European property market.
Richard Belgrave is LEVERTON’s Head of Europe, a role which means he is responsible for taking the company’s success in its native Germany and replicating it across the rest of the continent.
LEVERTON’s story starts in 2012, when the German Centre for Artificial Intelligence, recognised as one of the world’s “Centers of Excellence” in scientific research, created a commercial spinoff.
“Our CTO was a pioneer in neural network technology,” says Richard. “Which means, essentially, the deployment of algorithms that don’t only do what humans can do, but actually think like humans think.”
The LEVERTON business model really began to take shape when various large law firms in Germany caught wind of the fact that they could use technology to accelerate the process of reading contracts in order to speed up and improve real estate due diligence.
What and how is LEVERTON?
When it comes to machine learning, there are two common methods for reaching a point where a machine can think for itself. One point which Richard is key to point out is that LEVERTON have chosen the better option over the faster and easier option.
“I like to explain it this way;” says Richard, “A young child, maybe two years old, points at a chair and says, daddy, it’s a table. No, that’s a chair, you reply. Again, the next day, maybe in a shop, the child points at another chair and says, daddy, it’s a table. Once again, you say no, that’s a chair.
“The child has now been given two examples of what a chair is compared to a table. Next time, there’s a better chance they’ll get it right, and so the process continues.”
So LEVERTON’s method of machine learning mimics the way children learn, and although it is a slow and difficult process, it results in a far superior artificial intelligence than the other way of doing things, often referred to as rule-based learning.
“Rule-based learning is exactly what it sounds like. You sit the young child down and then attempt to explain to them every single rule about what makes a table, as well as the rules which differentiate tables from chairs.”
Richard explains that rule-based machine learning, although a popular choice of many ‘AI’ companies, falls short of perfection because all you are doing is enabling a machine to compare an object to the rules it has remembered.
LEVERTON’s method, on the other hand, doesn’t require the child to know all of the rules of what does and doesn’t make a table, they just need to be shown enough chairs and tables until they subconsciously, and instantly, know the difference.
You and I both know what a table looks like, but we’ve never been sat down and memorized the rules. We have simply learnt through experience.
“By the very nature of what we do, rule-based learning just doesn’t work. Contracts and other types of unstructured documentation don’t have any rules to teach, each one is structured differently with different jargon and discourse being used.”
He explains that, because it has been taught in this way, LEVERTON’s software is able to truly understand the language of contracts, leases, and deeds, etc.
“I’ve not seen anyone else in the market using machine learning this effectively for unstructured document review.”
90% of LEVERTON’s new business development is inbound, something which Richard puts partly down to PropTech and AI both being buzz-words at the moment, attracting a lot of interest from companies looking to future-proof their business models.
“It’s the promise of AI’s future potential that truly excites the property companies that I talk to. Yes, they all see the immediate power of LEVERTON, but when I explain the doors that increased AI sophistication is going to open moving forward, that’s when our clients get really excited.”
LEVERTON’s ability to expedite the deciphering of complex, unstructured documentation has applications for almost every industry you can think of. So why, as Richard says, is property ‘the only vertical LEVERTON has explored’?
“I think if you’d asked anyone 5 years ago, they’d have guessed that, by now, we’d be involved in other markets outside of property. But, as we’ve progressed, we’ve learnt that, despite it being a very high-value asset class, property’s perceived ‘norm’ is incredibly inefficient, not to mention it relies on the vast amount of unstructured documentation that we thrive on.”
During every meeting he attends and pitch that he gives to those working in and around the property industry, Richard is hearing the same thing.
“I soon figured it was better to stop talking and simply sit back and say, enough about us, tell us about you. Almost every time, they said that inefficiencies in contracts and documentation were a burden on their business. By learning about the client’s needs and concerns, we are able to offer real solutions.”
Giving sole focus to property doesn’t limit LEVERTON’s service to real estate firms. It’s for any business that has interests or investments in the built environment.
GVA and JLL are, however, their two biggest clients. When Richard first met with them, they had already done a competitor analysis and looked into building their own solution.
“Every company has a choice to buy or build. They both chose to buy.”
The reason they both chose to buy is that LEVERTON’s software is far superior to anything that either of them could build in any sort of affordable or timely manner. For example, LEVERTON isn’t only capable of analysing and extracting data from documents, it can also translate the whole lot into a language of your choosing.
This sort of innovation makes LEVERTON a joy for Richard to pitch, and is another example of very complicated technology delivering an easily understandable service with obvious application and value. It also helps catch the eye of some of the world’s leading property investors whose portfolios span the globe.
“We had a large British investor who came to us with a 4000-strong lease portfolio in 18 different languages. We helped create a structure by translating everything to English. At that point, they had for the first time, true transparency as to what their exposure in those countries was.”
Richard goes on to explain that the investor’s alternative was picking up the phone to all of those countries and speaking to hundreds of property managers, only to receive a load of inconsistent data as a result.
As well as JLL, Savills, GVA and Deutsche Bank among their impressive selection of FTSE 100 clients, LEVERTON also works with much smaller businesses.
“We have worked with clients with just eight leases at a time, and then others fifteen-hundred at a time; we don’t discriminate.
“We have a very fair pricing model and we don’t charge any set-up costs; we agree a price with each client depending on the amount of work that our system is going to need to do.”
Working with lawyers
Richard thinks a large catalyst for LEVERTON’s success is the fact that people are starting to wake up to the fact that lawyers are charging exorbitant fees for doing very non-value-added work.
“We do like to work with lawyers because they provide great value when you have actually got data; they can guide customers as to what to do with it. We don’t do any of that. We simply help you better understand the data that sits within a document; we bring it to light by processing it, extracting it, giving it structure, and putting it in the cloud.
“If you were to pay for a lawyer to do those things, they might take three hours to read a UK commercial lease, at a rate of at least £200 an hour. So, let’s say you’ve got 100 leases; that would probably take a lawyer about three weeks to deliver and cost you about sixty grand.
“For us, 100 leases would cost between eight and ten grand and be delivered within a week, maybe 72 hours.”
The lawyers are still needed to do the lawyering, but they’re no longer needed for the admin that they famously charge through the roof for.
“It’s a bit of a paradox for us, really; the law firms know that they need us in the long-term, but there’s a fear of being the first to move because then they’re effectively waving a white flag for their existing business model. They’re also reducing the fees that they can charge per transaction.
“And so there’s a large disparity in the legal world; some of them love us and want to work with us from the get-go, and others are saying, there’s no way this is going to fly; we have roofs to put over our heads.”
The latter are the ones most likely to struggle in the future. The arrival of AI signifies the dawn of a new industrial revolution, and while revolutions don’t happen overnight, it’s important that more senior partners find the courage to bite the bullet.
With such a simple and eloquent pitch and such obvious value, one would think that adopting
LEVERTON is somewhat of a no-brainer, and while their growth thus far has been impressive, the UK market has produced some unique obstacles, as well as some other less unique ones.
“To be perfectly honest,” says Richard, when I ask why people might reject his professional advances, “it’s inertia; the fear of the unknown.”
Richard says that if a company starts using AI, they actually have to start analysing all of the ways they currently do business.
“I see a lot of scared faces when I talk to companies. In one recent meeting that I had with a head of property management, after my pitch her face displayed a mixture of excitement genuine fear.
“She looked at me and explained how she currently employed seven members of staff to input information from paper leases into a property management system.
“So now she’s presented with a choice; does she go to the board and say that she’s found this tech company who can save them £250,000 a year in salaries alone, or does she feel such an affinity to her colleagues, many of whom she probably hired, that she chooses to turn her back on our tech? After all, the current system does work, it could just be done so much better. It’s a choice that lots of people struggle with.”
It’s easy to be sympathetic to that struggle, but sticking your head in the sand and pretending that nothing around you is changing has inevitably dire consequences. Educating people to appreciate this fact is the most common obstacle in property markets around the globe, but the way things are traditionally done here in UK also presents its own problems.
“The challenge in the UK is an interesting one”, says Richard. “There is this unique and strange dynamic of outsourcing. Companies don’t hire full-time employees to carry out non-core jobs. So when we speak to retailers, for example, they say something along the lines of, it’s really cool, what you’re doing, but CBRE already does this for us.
“For those companies, the idea of taking something that they currently outsource and therefore don’t have to worry about, and bringing it in-house, albeit with the convenience of advanced technology, is a huge psychological change.”
Richard notes that this problem is galvanised still by the fact that many of LEVERTON’s true benefits won’t be seen by their clients for a couple of years.
“To be honest,” he says, “some people just aren’t ambitious enough to invest and wait 18 months.”
Moving forward to the future
One of the most interesting things to consider about artificial intelligence is that its current level of sophistication is, over the coming years, going to be dwarfed. As the capabilities of AI increase, how will LEVERTON react, and how will their service change?
“For most of our clients, and most of the people I speak to in general, this is exactly what’s so exciting about LEVERTON.
“While we can already do great things, it’s over the coming years that the really incredible stuff will start happening. It’s the companies that recognise this that are most excited to partner with us. For me personally, the most exciting result of increased sophistication is our ability to scale.
“Our vision is to be a scalable software data business. At the moment, what we do requires a lot of human input and quality control. As such, our model is very much one of service delivery. When AI is sophisticated enough that we don’t need that human intervention anymore, and when the tech is so familiar that we don’t need teams to explain to clients how best to use it, our business can scale at a phenomenal rate.”
Looking to the future, Richard thinks there’s a lot to be excited about, and he sees LEVERTON a centre of it all.
“Big Data, for example, is a term which is still misinterpreted. People are spending all this time thinking and talking about Big Data, but in reality, they don’t even understand what their data is.
“We live in a world where all data is represented on paper, no matter who you are. Big Data promises to unearth that data and help you make better decisions, but because the process of extracting, reading, and understanding documents is so cumbersome and costly, we’ve found that most companies we speak to confess that they don’t actually understand their data in the first place, mainly because the data they have access to is of such low quality.
“LEVERTON will be able help people better understand their data, and soon people will realise that this structure and understanding is far more important that the concept of Big Data. There’s no point gathering as much as you can if you don’t know what it means or what it indicates.”
Looking into the future, does LEVERTON have plans to open up their service to industries outside of property?
“We will definitely enter more verticals. The property world will eventually move to smart contracts, and that pretty much kicks us out of play. We’re already speaking to banks and funds about the credit world but, to be honest, anywhere we see unstructured documentation, we will push our technology, focusing first on those industries where the underlying asset valuation is the highest because that’s where the contents of the documents in most important.
“Expect, by Q1 2018, to see us make a foray into a new vertical, but that doesn’t mean that property will be ignored. Leases are just the tip of property iceberg. There’s still a long way to go. That’s why any new vertical will be reliant on new members of staff coming on board; we’re not going to do it by taking our current staff’s attention away from property.”
Eventually, Richard hopes that people will see LEVERTON as a mini Land Registry; a source of truth.
“We currently promise not to sell or market the vast amount of data we collect, despite being regularly asked how much it might cost to buy. We already have, for example, expiry dates for 50,000 leases. The value of that data is unimaginable; we have to be very careful when deciding what to do with it.
“We’ve heard people talk about us as a potential unicorn. I don’t know how much truth there is in that, but I think that the data angle would certainly play a large part in any achievement of that sort.
“For example, when we have thousands of lease contracts in, let’s say, Norway, our data will be able to flag anything within a contract that doesn’t seem right, based on the dataset created by our portfolio of leases. The system will highlight the potential error and make sure nothing goes by unnoticed. The potential value of that alone, moving forward, is…well, again, it’s unimaginable; almost scary.“
Come on the journey with us
Despite all the evidence for the technology’s abilities, there are still AI sceptics out there, not least in property.
“I’ve heard things like, AI doesn’t work, or AI isn’t real. Some people even still believe that we employ 1000 people in India to do all of this, and then go around selling it as machine learning.
“I’ll be candid,” says Richard, “it’s always good to be candid. In the early days, we took on projects that we fell short on; under-delivered. But we went straight into this with clients who are among the biggest firms in the world. We didn’t start by dipping our toe, we dove straight in with JLL and Deutsche Bank.
“We learnt a lot, very quickly, about offering services and managing expectations of big clients. Sometimes, our service came back less than perfect, but that has really helped us to add safety devices to stop those things happening again. That’s the nature of machine learning, the more we do, whether results are positive or negative, it improves the service.
“But, early on, we probably didn’t have the right approach to client relations, so some of them soured. What we do now is start by explaining how we will take the client on a journey. We tell them what LEVERTON and AI can do right now but, more importantly, we also explain where the tech can take us all the near and mid future.
“As long as clients buy into you as a team, a company, a vision, and a person, they are understanding that the road we are travelling together is untested.
“What I say to anybody out there who is considering adopting AI for their contracts, is that it’s no secret that real estate is antiquated and stuck in its ways. But that also means there is an unbelievable opportunity waiting to be taken if people will open their minds to what AI can do for their business.
“Come on a journey with LEVERTON, and let us show you what AI can do, even if it’s just on a lease-by-lease basis; baby steps. Come on that journey with us, and let’s discover the future together.”
I have rarely spent time with a company in the property industry whose vision is as clear as LEVERTON’s. They know where they want to go and they know how to get there.
But that they also respect the fact that they are moving through unchartered waters, and that to presume that the future holds no surprises would be naive. That’s why they invite clients to cross the oceans with them, each supporting the other with a simple symbiosis that leads both parties to separate yet mutually beneficial goals.
LEVERTON might not be moving mountains, yet, but they are certainly helping move property safely into the future. For those that travel with them, the benefits are potentially too big to be measured, for those that choose not to, the consequences are simple.