AOCAP: Contract Review Automation Series: LegalSifter
(Republished from the Association of Contract Automation Professionals)
By Jonathan Drew
For our the second piece in our contract review automation deep dive, we spoke with Kevin Miller, the CEO of LegalSifter.
Kevin joined LegalSifter after nearly nine years at Industrial Scientific Corporation, a global safety company committed to ending death on the job by the end of the century. Kevin completed his tenure as Chief Operating Officer. Prior to ISC, he helped launch an online version of a regionally-accredited Argosy University, with bachelors', masters', and doctoral programs in business, education, and psychology. Kevin also spent nearly six years in a variety of roles at FreeMarkets and Ariba, its acquirer, in the global sourcing and procurement software space. Kevin holds a Juris Doctor degree from the University of Cincinnati College of Law, a Master of Business Administration degree from the Mason School of Business at the College of William & Mary, and a Bachelor of Arts degree in History from Davidson College. Kevin is also a licensed attorney in the state of Ohio.
Born of Carnegie Mellon University in Pittsburgh, Pennsylvania in 2013, LegalSifter was created with a vision- that thinking-reading-writing technology will unlock the potential of the legal profession. The platform focuses on contracts, using NLP (Natural Language Processing) and machine learning to bring a layer of ease and automation to the contracting process.
Thanks so much for your time today Kevin. To start off, I'd love to know a bit more about the history of LegalSifter.
LegalSifter has been around for a while, it got started out of Carnegie Mellon in 2013. I wasn't initially involved. All the local VCs were talking about natural language processing and machine learning, and how this new technology is going to automate routine and repetitive tasks. Well, there are a bunch repetitive and routine tasks in legal- and that was really it. For a year, LegalSifter sort of floundered around until they essentially stumbled into a use case. One of them had been a freelance software developer, and he had done some work for a company. Well, he delivered it, and didn't get paid. See, in the contract they had it said they would pay him upon acceptance, and they never accepted his work.
So the team went and said, "lets build something that reads contracts and gives advice for freelance software developers. And so they did that, and in august 2014 they launched the product. They got an article in TechCrunch, and over 5000 users in a few weeks. It was incredible. And then they put it on the shelf.
Apparently one of the lawyers that helped build this had gotten nervous, and the investors didn't really support the vision. Essentially, they didn't have the right leadership. Eventually, things spool back up, and in January of 2015, they start things back up again, and land a really big client - BNY Mellon, to do some extraction work- pulling information from over 25,000 contracts. BNY was doing everything manually, a big chunk of the work was being outsourced to India, so the team took the existing "sifters", the algorithms they built for the beta product, and put them into a user interface. And this allowed the team in India to get through contracts cheaper, faster, and better. So instead of 1 to 3 contracts an hour, they started going through 5-9 - huge productivity gains.
So tell me a bit about you- how do you fit into the story?
I was fortunate to land at a company called FreeMarkets straight out of law school and graduate school. That company went public in 1999, shot the moon and lives on today as part of Accenture and SAP / Ariba. After FreeMarkets/Ariba, I helped launch an online university for masters and doctoral programs. That business ended up being bought by Goldman Sachs. From there I went to a privately-held family-owned safety and technology company called Industrial Scientific. We had hardware and software services wrapped up in a subscription, and we had a SaaS business that had machine learning algorithms that predicted injuries in the workplace before they occurred. I served as Chief Operating Officer and spent almost 9 years there, helping to grow the business from $60 to $200M in revenues. It was a great ride. All told, I have spent a lot of time around contracts, both as a manager of legal, sales, procurement, and just about every other function in an enterprise. Contracts are a pain.
While I was in my last year at Industrial Scientific, I had coffee with an investor in LegalSifter. He said, "I've got this great company. It's got 4 employees. We have one client, and we haven't been paid in 2.5 years. We have no money, and no real product, but I think it's great."
That's quite the pitch.
My initial reaction was, I don't want any part of this -extracting data from contracts sounds super boring. But then they showed me the product, and talked about how it had been used for freelance software developers. Really what it offered at the time was a freelancer scorecard, it scored the contracts as well as gave advice for the freelancers, and the whole things took about 5 minutes to run. I saw this and immediately said "you've got something that reads contracts and gives advice? That's a big freaking deal".
So I decided to jump in. We bought out the second co-founder within 18 months. It took us two years to build the current iteration of LegalSifter, so during those two years we did extraction work to try to get a stronger sense of the marketplace, to get a real feel for things. When we launched LegalSifter we had 26 sifters - essentially 26 algorithms basically tailored to NDAs. Within our first week, our first client started using it, he's still with us today, by the way, and he said "I want to use this product on all of my contracts", which was a bit of a shock. We asked what he meant and he said "you know, I'm always reviewing indemnification. I'm always reviewing governing law. I'm always reviewing limitation of liability. And since you already have those sifters built, I can use this on everything." We had planned to move into other contracts types anyway, but it was really eye opening to see how fast some people got what we were doing and understood how broadly applicable it was. It happened much faster than we expected.
So you were sold.
For me, I just know this problem. I have managed lots and lots of salespeople, supply chain people, attorneys, etc. and they all complain about each other during the contract process. And the issue is that they don't have the technology needed to facilitate a good relationship. Getting the tacit knowledge out of people's brains and having it react to the process, a process that's a little different for each transaction. You could have all the databases you want, all the templates in the world, and every workflow enablement tool out there, but the biggest issues are the negotiations, which slow down cash flow, and inserts risk in the form of bad negotiations. And I know this implicitly because I've felt the pain. So our strategy is to sit right between the service providers, the lawyers and consultants out there, and solve the biggest problem that people have. Make sense?
Absolutely. So tell me more about this problem, and how LegalSifter is solving it.
The primary problem with contracts that is largely unaddressed because of gaps in technology, or even in strategic understanding, is that in every contract negotiation in the world, at least one part is looking at a contract they have not created, they are looking at another party's paper. So we built a product that reads contracts and gives advice. Its available out of the box. If you want, you can configure it. If you're a law firm, you can configure it to your specific guidelines. It's called LegalSifter, and the intent is that for every contract negotiation, it operates like a spell check or grammar check. You upload a contract, select what type of contract it is, which tell us which sifters to use, it tells you what's there, what isn't there, and what should be there. That's what it does.
The benefits of this are multiple- number one, it makes you individually quicker. Think of it like Google Maps, right? I know how to get to my kid's school, but anywhere else in Pittsburgh, I'd better have Google Maps on. Overall you're seeing a 20-80% time savings.
Second thing is that contracting is a multi-step process. If you're a large organization, then you get compression on the actual process itself. So if it's a three step process, every step in that process is now a little easier. So what happens is you end up squishing the duration, meaning faster time to cash on the sell side, and you're also lowering risk. So I'm signing contracts with better advice, taking that advice more frequently, meaning the company is getting more consistent. At the end of the day, this also means I'm getting better outcomes on my negotiations as well. The idea is that in the not too distant future, all of us will use a tool that works just like a grammar check, it will be that easy. In all of our simple and complex document review, NLP and machine learning will have in context advice, and everyone will use it. We are a "concept" check. We make sure that all the relevant pieces are there, and that things you don't want there aren't there. And it's as easy to use as grammar check.
So as easy to use as grammar check, and equally ubiquitous. Tell me a bit about how people are using LegalSifter today.
Think of it like this. Lawyers sit there every day, and they have a few simple use cases. First, they review another party's paper. They double check their first draft, then they send it over to you to double check it. So now that markup is being reviewed. Now maybe I'm at a larger organization, so now we're trying to get int on our paper. For large law firms, you're always checking your teammates' work. First and foremost, we're automating that.
Interesting. So if I'm understanding correctly, for large clients or law firms, they will have the sifters configured for their specific language. But if say, I own a bike shop, I can use LegalSifter to make sure the contracts I'm signing are sound, without needing to be a lawyer?
Exactly. It's not just making lawyers' lives easier. It's taking the idea of simple contract management and making it readily handleable by non-lawyers. And you understand this because most contracts on this planet or negotiated by non-lawyers, that's where the pain is.
We all think our lawyers are too expensive and too slow which they are. But it's not their fault. Lawyers don't have technology that makes them cheaper, faster, safer. That's because they needed this technology, NLP and machine learning, to reach to the novel language and documents they encounter on a daily basis. That's why this technology is so magical for this particular problem- because contracts have novel language, every one of them. So because you have to react to this new language, the old keyword strategy doesn't work. Now the technology lawyers need has arrived.
But the technology by itself isn't enough right? People have to get it. People buy LegalSifter for fast advice. That's what we're providing, and the results are the other benefits. So the fast comes from the user interface and the NLP and machine learning, and the advice comes from humans who've set it up, and that's why it works.
Last question – what do you think the future of legal is going to look like?
Our mission is to bring affordable legal services to the world by empowering people where they are. The fundamental challenge people have with attorneys is that they are too slow and too expensive. People want to blame billable hours, or a disinterest in technology, well the reason they haven't been huge adopters of technology is that we haven't been giving them technology that does what they do or really fundamentally changes the way they do their job. You can yell at lawyers all you want, but at the end of the day most technology being marketed or sold to lawyers lives on the periphery of what they do.
One of our team members is a former 30-year partner from a large law firm who came to us at the tail end of his career. He told us legal technology, for his entire career, has been like a doughnut. He operates his job in the center of that doughnut, and the technology is all at the periphery. It might help with research, which you do once in a blue moon, or help with billing, or project management, but it doesn't help me do the core job. Because at the end of the day, in legal, its all about the lawyer. The lawyers experience, the lawyer typing. This is obviously less true with discovery and litigation – technology has already been of incredible value there, but when you sit down and look at it, and compare legal to other functions and other industries, there hasn't been the same scope of innovation.
If you read all the blogs, all the articles, people love saying lawyers hate technology. But I think it's the opposite- lawyers are all using technology, smartphones, smart homes, smart cars, its just that no one has made technology that solves lawyers core problems until now. Now, with NLP and machine learning, we can finally give lawyers the tools they've been asking for.