The AI Lawyer | Jerry Ting from Evisort

DISCLAIMER: Below is an AI generated transcript. There could be a few typos but it should be at least 90% accurate. Watch video or listen to the podcast for the full experience!

Jerry: We’re purpose built. And what I mean by purpose built is we actually sat down with a team of data scientists. And we looked at the problem. And we said, how do we solve this problem from a first principles perspective?

Alexander: Welcome to UpTech report. This is our Applied Tech series UpTech Report report is sponsored by TeraLeap, learn how to leverage the power of video at TeraLeap.io. Today, I’m joined by my guest, Jerry Ting, who’s based in San Francisco. He’s the co founder and CEO of Evisort. Welcome Jerry could have you on, man.

Alex good to be on. Now Evisort, your platform as states in your site is an AI powered contract management platform. Basically helping those companies out there, organize and figure out what is all that important data contained in your legal documents. your target market, those who you’re helping is definitely enterprise and upper mid market. But help me understand the the intelligent contract management platform. That’s what that’s what you state. And in our conversations before our interview here, we were talking Okay, what is AI? I mean, what how can you really say what is AI is just a marketing term. Tell me what what what did you go into building Evisort? What was the mindset you had and the problem you’re trying to solve?

Jerry: Yeah, I never wanted to start a company. That’s the way to start today. Yeah, I was gonna be a lawyer. And I think that’s sort of what resonates with, you know, what is real AI versus what is fake AI’s. For me, I’m, I’m a straight shooter. I’m black text law, if you will. And so for me, I was in law school. And I was studying to become a lawyer, I wanted to be a startup lawyer, I wanted to work with folks who work with technology. And I quickly learned that the practice of law was so manual. And what you have to do is, let’s say you’re negotiating a lease for your apartment, you sit down as a lawyer, and you open a 200 page document and you’re reading, can you put a sticky note on the on the wall, or if it takes off some paint, you can get in trouble, right. And it’s so manual, and I, and I sat there as a lawyer. And I said, Wow, this is a very tedious task, and also is a very expensive task. I went to law school was going to work in a big law firm in New York, and you’re going to charge people hundreds of dollars an hour to, to, to review documents. And so I thought, Hey, there has to be a better way of doing this. And I was volunteering actually, in the Boston community, to help founders start companies. And one of my clients was actually a researcher at MIT. And he was actually working on AI research. And so I went to him. And I said, Look, I know you go to MIT, clearly, you’re pretty caught up with the technology. What is AI, you know, AI can actually help lawyers do their work. And that’s how the company got started was based on my personal experience, being a lawyer, and reviewing contracts and helping my clients. And my client and I actually got together and started building a company to solve that cause. And so that’s right, you know, getting

Alexander: teaming up with a client where you who has the understands the technology, you see the need, and you’re like, I don’t want to do all this manual labor that just waste people’s money and my time. And that’s where, hey, we’d like technology, we can automate and simplify our lives, we can focus on what what we what we can do better. But help me understand let’s let’s take a little bit further. Okay, what is true, true AI? And there’s, you had mentioned earlier in our conversations about there’s there’s legacy players that come in, come in here and then slap on maybe some regression systems or whatever. And it’s 40%. Accurate? And then say, we’ve got AI. Okay, what’s the difference? Like? How can you tell what is true? Ai? Like, who defines the term?

Jerry: Yeah, it’s not a regulated term. So it’s, it’s whatever your marketing team wants to define as AI, which I think hurts consumers. Because as a CIO, for us, we sell to enterprises, and we sell the mid market clients. For some of them, it’s the first time that they’ve ever seen a intelligence system like this before. And so for them, they are looking at maybe four or five different systems, side by side. And for us, we’re, we’re purpose built. And what I mean by purpose built is we actually sat down with a team of data scientists. And we looked at the problem. And we said, how do we solve this problem from a first principles perspective? How do we actually use math? And how do we use deep learning in machine learning? And for us, because we’re looking at language? How do you use natural language processing, which is, which is a part of AI? How do you use that field to do acoustical analysis, right? It’s a totally different approach than if it’s a nervous system that is more of a database system. And then they put on some role based approaches on top I’ll give you a simple example. Sometimes you will get purchase orders. And one of the things our clients need us to take out of a purchase order is the name of the vendor that you’re paying. simple problem, right? For a lot of our clients, actually their purchase order is standardized. It’s on their form. So their accounting department issues, the purchase order, the procurement department issues, the purchase order. In on the top left hand side, it will say the name of the vendor. Let’s say the name of the vendor is Salesforce. So this is vendor, a colon, Salesforce, anybody can pull that out even a non AI vendor, right? Because you just say look to the top left, and you just set a set of business logic stack and look at the top left. What happens if it goes and moves to the top right? Why does it it goes to the 14th page and is buried in a paragraph. That’s when you separate the real UI vendors from the quazi advice from

Alexander: the men. Exam when it comes to AI. interesting analogy is helpful because there are I’ve talked to a few vendors where they talk about Well, yeah, it as long as it’s always in the same place, it’ll be able to pick it out. But if it’s all it could be different each time. That’s you’re saying is true AI. So you believe that there should be more regulation on on being able to use the term AI?

Jerry: I do. I do. Because I think for us right now, we’re still in the nascent sea of AI and its maturity and its application due society. So you know, AI has been talked about for the last 10 years or so really, really, it’s been the last five years where we seen a real impact on our human life. I drive a Tesla, for example, right, and I rely on the car to make some decisions on the freeway, I’m still president, my hand is still in the wheel. But when it’s going around a term, I put my life at stake going 65 to 70 miles an hour, rely on camera sensors and an AI based system. And so I think as AI becomes more prevalent in our life, it is no longer a academic term that folks like you and I who are board nerds on the topic that we care about. Right impacts our parents and impacts our family. And I think it impacts how we can work. And so for for that reason, I think having some regulation around how you define AI, and how you talk about accuracy and safety. I think those are important areas where you got to draw a line in the sand. Where is it actually accurate? Or is it just a legacy system? That’s a rule based logic pretending to be an AI system. And I think the purpose of that regulation is the guide consumers so that they know how to rely on the system, and how not to rely on the system.

Alexander: Until regulation appears. So eventually, I imagined it well. How can a company a business leader themselves be able to compare apples to apples or rather apples to oranges here? of Okay, is this truly an AI solution? I mean, what are some some some giveaways?

Jerry: Yeah, I, one of my first clients, she she taught me a phrase that I never forgot. She was General Counsel of a company. And she said, I like you when we do the Missouri way. And I said, What’s the Missouri way? And she says, Well, that’s the show me state. And, and what she meant by that is, I’m going to give you five contracts that you’ve never seen before, scanned from different systems written by different lawyers. You say you’re good at pulling data out of contracts? Let’s upload it. And let’s see. Right. And when I look at the results, there’s there’s no funny business. Let’s don’t don’t leave the room. I’ll put it right here. And let’s take a look at the results. And I think that that was a really good learning lesson for us, because we started telling other clients, hey, there are people who say to do AI Sure, there’s, there’s what are the only ones where there’s others out there. But there’s not 200 there’s only a couple that really do it. And and so what I encourage you is whether you work with us or don’t do them a certain way, upload some data. Let’s test it right, let’s see what it actually does. And for me from a personal perspective, when I first started driving my car, I was really nervous. I gotta be honest with you, Alex, I was driving and I was I was holding on, I was more nervous with the AI system. In the test. I did that when I didn’t have any AI system because I didn’t know Am I going to crash into a wall or not. But over time, as I went through turns as I went through different sections, the highway, I realized that this system is pretty accurate. And then that’s what I was able to kind of relax a little bit and let the system do what is meant to do. Same with using a sword. Same with using any other system is that you got to let it prove itself. And if it can’t prove itself, then maybe there’s some smoke and mirrors happening her

Alexander: It’s a powerful thing to show me take a ticket for a test drive, so to speak, or literal if you’re actually driving a Tesla. Some giveaway something you talked about also is You know, when you look at the founders, the people who start an AI or ml company, do they have a PhD in math? And like, help me understand, like, why do you You said that earlier before we start the interview? Why did you say that?

Jerry: Yeah, I think it’s because AI is still a new field. For us. You know, my, my co founder was doing research at MIT, he was actually part of the academic research groups, or writing papers, contributing to research. And I think it’s a different type of thinking, when you’re aesthetician or mathematician, or data scientists, versus when you are a software developer, focusing on building software. Not to say one is better or worse than the other it just different, right? Well, one’s a tennis player, one’s a basketball player. And it’s sometimes you can cross over, but it’s harder, right? What has a racket and one as a one as basketball. And so for me, because AI is so new, right? It’s there’s been a lot of hype around generalizable AI, you know, big tech companies, including some that we partner with, talk about building AI research that you could just as a software developer, take off the shelf, aim at a set of problems and then build AI magically, right. In our experience, in our testing, that has not been the case, when you take, I won’t say names, the big tech companies, because we’re, we’re partners with almost all of them. But when you take something that’s not built for a particular set of challenges, you take it off the shelf, and then you just give us some training data, as a software developer, and then you test it, the accuracy rates are all over the place, right? It works really well on certain things, works quite poorly on others. And so for us, I think, you know, I speak with a lot of folks who run AI focused companies, we call it an applied AI, I think you and I talked about, what is academic AI versus what’s applied AI. For us, you know, our AI was built from the ground up, looking at contracts, right? I’m a lawyer, I’m a subject matter expert. I’m working with a data scientist who’s actually inventing algorithms. And we’re doing live testing side by side, right at Intel, the thing passes the human eye test, which is the most important test in AI. It doesn’t leave the lap. And I think that that’s a really important way of thinking about.

Alexander: Another topic we talked about earlier. And I want to dig into here is our automation versus augmented an augmentation of like, how much do you automate something versus someone is involved in this? Again, we can go back to the car analogy, your Tesla imagine? To what degree do you feel comfortable and safe to let it just do its thing? versus wind? Do you feel okay, a person needs to be involved in here, and especially when it comes to contracts? Where where’s the balance? Where’s the handoff?

Jerry: Yeah, I always think that there’s going to be a role for humans in in an AI based society. I’m a lawyer, there’s things I learned in law school where there’s no way I will know that, you know, the, I was lucky to be taught by folks. I went to I went to Harvard for law school. And my first year contracts Professor thought Obama, and the lessons that he learned in that in the classroom, in AI will never know, right? And when you get into decision making formats, where you have to think back to the human components and making decisions in AI, whenever No, right? Yeah, and for me, I have so much respect for lawyers because of what they do on a day to day basis. There’s also things they do on a day to day basis that they don’t want to be doing and they are doing right and, and we I laugh a lot when I when I hang out my friends at the bar after work. And I say, Well, do you want to be doing that? They said, geez, no, get me out of here, right. And so for me, the role of augmentation versus automation, is automation is better for simple repetitive tasks that don’t require decision making. An example might be you go to a vending machine, you put in $1 bill, and a coke comes out. That’s full automation. You don’t need to go in there and check Hey, did you choose the right bottle of coke? Or did you choose a sprite there is like a rule based logic that Coke is in the same canister every time. Right? And that’s good for automation. When you do still driving, or you’re reviewing a contract that has $100 million attached that it’s worth having a human check the output sometimes when when the system is doing complicated decision making, like for example, when I’m driving, right, that you are Musk is a visionary. I look up to his his his vigor and his work. But he never says go to sleep when you’re driving a Tesla, because somebody might die. And I think that that’s an area where augmentation is better than automation, and consulting where I had some background. There’s a rule called the 8020 rule.

How do you get rid of how do you focus on 80% of value, and then don’t forget it. You just kind of let go 20% goes to the wayside, right? But don’t solve for 100%. If you solve for 100%, every time you’re not going to get out of the garage. And so for me, I think when you’re designing systems based automated intelligence systems, design principle that we use is how do we get rid of the 80%. So that you can focus on the 20%, actually, which is where human decision making is very relevant. And I think that’s a difference between automation and augmentation. And in my perspective, I think augmentation is just as effective as automation. For real life applications of AI.

Alexander: You mentioned earlier that you have several partners and partnerships, what’s what’s your integrations look like? Obviously, with contracts, it’s important that there’s a flow of information always coming in and going out. What type of the things that have you found that are integrating with and are planning on?

Jerry: Yeah, this is an area where I think you hear the phrase SAS, and then you hear this phrase AI. And then you say, is AI part of SAS or SAS or AI? And I think it’s actually a that’s that’s a longer debate. But for us, when I go into client organizations, when we we have so many clients who use Salesforce, they use Reba, they use Tableau and Power BI visualization, they have accounting systems, they have workday, right? If you think about the last 20 years of software development and the migration from on premise to cloud, right, that’s sort of the sass generation, if you will, I think we’re at the tail end of that. I think we’re entering now the AI plus SAS generation, where systems are now naturally intelligent, right? You think about a system like Oreos, you think about systems like we use gotten, for example, for sales recordings, it actually tells you, hey, what should he say? What should you not sing? Right? So I think our systems are now moving on towards more intelligent system. So to your point, it’s really important for a vendor like us to integrate with the sales forces, the Microsoft’s the tableaus of the world, because that’s where business is currently done today, because those are the incumbent players. And for us, that’s when data lives, a lot of data is manually entered into these other systems, what are the downstream systems, or upstream systems, and for us to be an intelligent node that connects these different systems, we actually make these other systems better, because we can actually supply data into your Salesforce to help your sales teams have more information, and to help your sales teams actually close deals more quickly. And I think that that’s a very interesting synergy, right? We call it frenemies, if you will, friends and enemies, right. You got to play well with other folks that you’re that you’re integrating with. But sometimes you’re going to have overlapping value props, where you say, hey, I want to own this value prop. I don’t want to integrate that with a with with a sass system. I think that’s a ever evolving synergy that I think actually helps consumers, because you’re actually expanding the pie with what’s valuable to them.

Alexander: What can you share of your roadmap, what’s coming up? What’s exciting feature that is just come out or will be coming out that you can talk about?

Jerry: Yeah, I’m gonna get in trouble with my r&d team. But I’ll go ahead and go ahead and talk about it. For us, there is always been always been a gap in the legal technology industry, of canny AI actually negotiate a contract. We can now we’re at the stage now where we can pull data out, we can tell you what it says we can tell you, if you put in a contract, who’s a with when does it expire? What are the key risks we can tell you about already? That wasn’t possible five years ago? The next step is can I use an AI to negotiate against a human. And that’s where it starts to get really interesting, this line of human intelligence versus artificial intelligence. I’ll say here that we’ve actually started already doing this, one of the biggest banks in the world, where they are negotiating 400 page contracts, using eversource as a part of the negotiation framework, where we can actually look at the comments that the other party made with other parties and bank or an asset manager. And this, this large bank sends out a contract, it comes back negotiating with run lines, where the counterparty has some edits to make, we can actually look at those edits, we can tell you, Hey, does it comply with your contracting policies? Or does it not? And if it doesn’t, what should you ask for instead? And I think that that’s really interesting, because that’s kind of like the Tesla driving by itself with you supervising it. We’re going to be releasing that to all of our clients in the next 12 months.

Alexander: So basically, as the contracts come in, it’s highlighting with changes that they’ve asked for and giving you already suggestions of based off of what you say you want. You could reply with this, doing all that for you with you then being able to reply manually.

Jerry: Exactly.

Alexander: Well, that’s that is that is exciting is in that direction of automated pilot, Tesla mindsets. When it comes to contracts. I, I like the future that you paint that it’s applied AI, like it’s something that actually is making our lives better. But there’s still a lot of room for growth in the direction of where where things are headed. For those that are involved with legal documents, whether it’s in marketing, procurement, sales, more legal, obviously, is there any words of advice that you would give to a business leader in any of these departments? When it comes to contracts and the future of how to manage them?

Jerry: Yeah, I think for it, if you think about it, anyone that touches a contract, cares about the data inside of a contract. I’ve never gone into a company and they said, We don’t care what’s in our contracts, has never has never happened, right? Because every time a company buy something, or sell something, or hire somebody, there’s a paper trail, that data is very valuable that data could I think, move stock markets. And so I think for, for folks who are thinking about managing data instead of contracts, if you find yourself or your teams or your partners and your other departments that you may work with manually reading contracts, and then typing data into a database, you got I urge you to stop and think is that the best use of time? Or should we think about how do we actually augment that with an AI based system, so that your teams are actually looking at the data, and then maybe making decisions with that data? I’ll give an example. We, this week actually met with the chief procurement officer of a fortune 100 company. And the officer said, I want to take out over $100 million in savings in the next couple of years. And they asked me, How do you do that? And I said, that’s actually a sounds lofty, but it’s actually a data question. It’s not a question about a strategy that procurement plays that, you know, everyone else knows other people know how to how to do this. But the question is, how do you get the data? Right? There’s three main use cases in procurement, where you can use data to drive cost savings. One is managing your contracts, and your explorations effectively. If you don’t want to service, you don’t want a piece of software, and you miss the expiration dates. You’re on the hook for another 12 months, right? That could be $100,000. That could be a million dollars. Or worse, if you miss the expiration date, depending on the terms of the contract, you may be on the hook for another 30 years. And and so let’s at least get that house cleaned up. Right, that’s that’s the first bucket. The next bucket. There, there’s a big trend in procurement right now to consolidate spending, you know, SAS was such a prevalent growth space where, you know, you take a big company, like a, like a fortune 100, you know, different managers, maybe all bought different licenses and Salesforce, because they needed for different use cases, marketing media for marketing, sales needed over sales, different regions of sales, North America versus APAC versus amea. Do maybe 10 different licenses to Salesforce, all with no discounts. As a procurement officer, you want to go across your entire company and figure out, Hey, can we get one license? And can we negotiate 30% off, there’s another couple million dollars in savings, right. And then the third bucket just I’m just illustrating a point here of how data can be used to drive business. Now another third bucket is there’s rebate tables. In vendor contracts you you have rebate tables, where if you’re a large company, you spend you know a million dollars with those, you get back $50,000. Right? If you spent 1,000,001 point 5 million you get back $100,000, there’s like there’s literally a table of if you spend so much compared with how much you spent in the APR system, and you can actually collect the rebate. companies usually don’t do that. They negotiate for rebates, and then they never have to go and collect the rebates. Another couple of millions of dollars, you can see how when you’re looking at data that’s basically dollars are represented in text, how you analyze that text, you can drive, really large savings and efficiencies for companies. And that’s what gets me excited about AI. It’s not the it’s not a math, it’s not the Sigma notations, that my data scientists are putting on whiteboards. I don’t really understand those. And to be honest, they draw the most fancy graphs, right. But for me, if you if you take that and you don’t apply it to a business challenge, you’re missing the point. It’s about how do you take best in class r&d, apply it to a business challenge that is very valuable, very legacy has been done manually. We’re not done at all. And when you combine the two, I think you see the future. Yeah, I think that that’s the direction the world is going right now.

Alexander: I love the future that you paint and I am 100% with you all Applied solutions. It’s wonderful theoretical and research, because that paves the way for things but when they actually get applied, that’s when life ideally becomes easier and people’s lives can can be more focused on things that they can do better and themselves. Thanks again, Jerry for this advice for those that want to learn more about Evisort you can go to EVISORT.COM, and be able to get a demo there. And for those who want to hear more about Jerry’s journey and some of the insights he’s had over the past few years building Evisort, stick around for part two for our founders journey interview series. Thanks again, Jerry and everyone will see you on the next episode. That concludes the audio version of this episode. To see the original and more visit our apptech report YouTube channel. If you know a tech company, we should interview you can nominate them at Uptechreport.com. Or if you just prefer to listen, make sure you’re subscribed to this series on Apple podcasts, Spotify or your favorite podcasting app.