New
Patrick Ip: How Theo Ai Is Redefining Legal Predictions
This week, we kick off the new year with Patrick Ip, co-founder and CEO of Theo AI. Patrick joins the podcast to discuss his journey from Google to entrepreneurship and how his company is leveraging AI to transform legal workflows. As the legal industry begins to embrace AI, Patrick shares his unique perspective on opportunities, challenges, and the ethical considerations surrounding these groundbreaking technologies.
The conversation begins with a fascinating discussion about a recent pro se lawsuit where AI tools like OpenAI’s GPT-4 and others played a pivotal role in drafting a complex complaint. Patrick and the hosts delve into the implications of this case for legal professionals, highlighting the advancements in AI’s capabilities and the need for caution when non-experts wield these tools. The discussion provides a critical lens on the ethics, risks, and reliability of integrating AI into the legal process.
Patrick shares the inspiring backstory of Theo AI, rooted in his rich professional journey, which spans work at the United Nations, launching startups, and being part of a Nobel Peace Prize-nominated project at Google. At Theo AI, Patrick has combined his entrepreneurial spirit with his legal expertise to develop tools that make legal predictions more accessible and reliable. From managing client expectations to transforming litigation funding, Theo AI’s innovative use of synthetic and firm-level data is driving efficiencies and fostering better decision-making across the legal landscape.
The discussion also ventures into the practical applications of Theo AI, particularly for litigation funders and law firms. Patrick explains how Theo AI compresses case review time from weeks to mere minutes, offering predictive insights that help legal professionals assess case viability, manage risk, and optimize workflows. He emphasizes the role of trust and transparency in AI development, ensuring the technology is both robust and aligned with ethical practices.
As the episode concludes, Patrick reflects on the future of AI in the legal industry, forecasting that the most transformative advancements will seamlessly integrate into existing tools like Microsoft Word and Outlook. He also shares his broader philosophy of balancing work with personal passions, drawing inspiration from his experiences as an entrepreneur, coffee aficionado, and triathlete. This engaging conversation is a must-listen for anyone interested in the evolving role of AI in legal technology and beyond.
Links:
Listen on mobile platforms: Apple Podcasts | Spotify | YouTube
Blue Sky: @glambertpod @marlgeb
Email: geekinreviewpodcast@gmail.com
Music: Jerry David DeCicca
TRANSCRIPT
Marlene Gebauer (00:05)
Welcome to The Geek and Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gabauer.
Greg Lambert (00:12)
And I’m Greg Lambert and on this week’s show we are joined by Patrick Ip, co-founder and CEO of Theo AI. Patrick, really happy to have you here on the Geek in Review.
Marlene Gebauer (00:24)
Welcome, Patrick.
Patrick Ip (00:25)
Yeah,
thank you for having me and happy new year.
Marlene Gebauer (00:27)
Happy New Year. Speaking of the New Year.
Greg Lambert (00:28)
Happy new year, wow. Yeah, 2025 doesn’t
even feel real to me, I’m not sure. Yeah, the 80s were still like 20 years ago, right? So I wanted to kick this off and have a little side discussion before we jump in and talk about your work there at Theo AI.
Patrick Ip (00:35)
We’re still in 2020.
Marlene Gebauer (00:35)
It’s just an extension of last year. Exactly.
Greg Lambert (00:53)
And there was a really interesting case that was filed in the middle district of Pennsylvania last week. And two brothers and these guys have posted things like on Reddit to explain what they did. But basically they had a fraud case. I think there was some crypto involved in, but they’re suing, yeah, they’re suing for something like a little over $26 million.
Marlene Gebauer (01:15)
It’s a digital currency group, yeah.
Greg Lambert (01:22)
But they decided just to take this case up on their own with the help of tools like OpenAI o1 Pro, which is the new $200 a month version. think they also used Gemini and Claude to help them draft a 100 plus page complaint that they filed. Now I saw this and…
Patrick Ip (01:40)
Thanks
Greg Lambert (01:50)
Something cued me in on one of the social networks and I downloaded it and I took their complaint and I put it through Shepard’s AI tool that I have at work and I only tested it to see if it got any cases that it cited. Did it get those cases right? Did it hallucinate? And if they quoted those cases, did it get the quotes right? And it actually passed with flying colors. So I think, you know, we’re
were one step up from where we were before. Yeah. It got the citations right. However, apparently, as our friend, Joe Patrice said, Above the Law said, when he looked at it, that although it looks good, it was about 98%, 98 point something percent really good. But that’s also the difference between the DNA between apes and humans.
Patrick Ip (02:22)
Yeah, definitely an improvement.
Marlene Gebauer (02:23)
At least it got the citations right. But…
Greg Lambert (02:48)
And so, know, it’s a little bit and he referred to it and I think this is a term of art for the law as “Hot Garbage” And so I just wondered, did you guys get a chance to look at that or what’s your thoughts of someone, you know, totally relying upon AI right now to file these types of complaints with the courts?
Patrick Ip (02:58)
Hehehehe
Yeah, you we’ve talked about it within our company too. And obviously this is not the only case, I think a year or two ago, right? People were making up fake citations or know, ChatGPT was. And I think maybe my different take on it is the technology has exponentially gotten better. You know, the o1 model for the first time it can think, which is slightly different than what the three five model did. But I think the main thing that I think is different is
One of the things about ChatGPT we’re giving access to any person that’s non-technical to work with an AI model. It would be the equivalent of kind of giving a Ferrari or F1 car to the average driver and say, hey, drive it in a straight line, make right turns, make left turns. There’s a high crash rate when you have a non-technical person using a supercomputer. But if you give that technology to, on our team, we have, for example, this guy named
Dr. Alex Liu, he was a former chief scientist at IBM. He did prediction models for NASA. And when he goes and uses the o1 model the things that he produces are absolutely astronomical and amazing. And so I think you’re going to see maybe a difference in output depending on who the user of this technology is. I think as a technologist myself, I can attest that think these models have taken a leap and we’re going to see some amazing things this year.
coming out from AI legal tech companies. But I think part of the reason that we see these headlines around these models aren’t very good, they hallucinate, that’s just for anyone that’s logging into the o1 model, they’re using a consumer version of the application. But if you talk to any AI person, none of us are using the model that you’re using. We’re doing a whole bunch more on top of that.
Greg Lambert (05:06)
Yeah. Yeah. Marlene, what are your thoughts on this?
Marlene Gebauer (05:06)
Well, you
know, I would say that, you okay, maybe I got the citations correct, but…
You know, you’re also thinking about, you know, authenticity of the claims, know, accuracy of the claims. I think I think there was something on Blue Sky that a lawyer said that the brothers actually pled that their LLC is a fraud and a sham. And, know, and then there’s the ethics of it. You know, there’s you know, that’s questionable in terms of, you know, filing something that I mean, they are pro se, but, you know, filing something that, you know, you really have no idea whether it’s
Patrick Ip (05:30)
Haha
Greg Lambert (05:31)
Yeah, yeah, I saw that.
Marlene Gebauer (05:45)
good
or bad. you know I would say that yeah that it’s problematic that if people start getting that idea that I’m going to try that and just see how it happens because again maybe these guys are fine and you know but if they get slapped with sort of a counter suit maybe they can handle maybe they can handle that but a lot of people in the real world can’t if they basically put something together that is as as Joe said “Hot Garbage.
Greg Lambert (06:11)
Yeah,
now I did want to point out that I talked with a litigator at my firm and a well experienced one and he had somewhat different, I think he was a little more…
little less critical of the brothers. And what he said is, you know, a complaint is really like the most basic thing that you can file to the court. That you’re essentially just giving a statement of facts in order to raise to it, to make a claim. And so he thought, you know, hey, even though this one, because it’s a pretty complex
Marlene Gebauer (06:45)
Maybe.
Greg Lambert (06:57)
issue I think with this type of fraud that you know he was a little more sympathetic to them and thought you know if you can get the AI to do that, it more power to you but if you go beyond this you’re going to need some serious help and his suggestion was you know for $26 million I’m pretty sure you can find someone to take this on a contingency basis.
Marlene Gebauer (07:25)
For
$26 million, people might be like, hmm, if things are not correct in that complaint and they open themselves up to a counterclaim, I think maybe frivolous lawsuit, the whole thing. I do think there’s some danger in filing a complaint like this.
Greg Lambert (07:36)
Well, it’s. Yeah, what?
Yeah, I
think maybe the horse has left the barn on getting someone to take it on a contingency, but you never know. There’s a pretty large amount of lawyers out there that might take a gamble on it. So, all right, well, I think we’ve beat that one to death. Let’s get into why we actually brought Patrick onto the show.
Marlene Gebauer (07:48)
Yeah, I think they’re done.
Patrick Ip (07:48)
Too late now.
Marlene Gebauer (08:03)
Yes, so Patrick, you have quite an impressive background. You’re an ex-Googler, a serial entrepreneur, and you’ve been featured in Forbes 30 Under 30. And you’re even part of a Nobel Peace Prize nomination. So can you tell us a bit about your journey from Google to founding Theo AI? I don’t think we’ve ever had a Nobel Peace Prize nominee on. I think this is a first.
Greg Lambert (08:22)
Yeah
Patrick Ip (08:23)
Hahaha.
Yeah, I can give you the of the backstory. It starts a little further back. So when I was an undergrad at the University of Chicago, you know, I was obviously an ambitious student and I wanted to know what the most successful people were doing. And I was told as a freshman in college that the most successful individuals were management consultants and investment bankers. And so as a freshman, I applied to JP Morgan and Goldman Sachs and McKinsey But of course, as a freshman, I got rejected. So was like, oh no, what am I supposed to do for my freshman year internship?
And I happened to read this article about this foundation in Melbourne, Australia that was working in education reform. And so I cold emailed the CEO and I was like, Hey, I’d to come work for you. And long story short, he pays for my plane ticket from Chicago to Melbourne, pays for my apartment during the summer and gives me a summer stipend. But then two weeks into my internship, he resigns as a CEO, but he also resigns from his post at the United Nations. And he’s like, Patrick, do you want to take over for me at the UN? I’m like,
Of course. So at the age of 19, I get elevated to senior official status of the UN of Australia. This is also my first time stepping foot in Australia and suddenly I have this post. But also as the youngest person there, they’re like, hey, you’re 19, you must know something about social media. Can you open up the social media accounts for the UN? I’m like, sure. So I opened up the Facebook and Twitter accounts. It just so happened that Melbourne was also hosting the 63rd annual UN conference.
Greg Lambert (09:27)
yeah, we wouldn’t.
Patrick Ip (09:54)
And so during that conference, we got 1.2 million hits, which back in 2010 was a decent number. I would basically go on and run social media for the UN for the next couple of years, which led me to start my first company, which is a company called Kip Solutions. It was a social media consulting firm for not-for-profits. And then my senior year of college, that business got acquired by a consultant agency in New York called Post and Beam, moved to New York after school.
did that for a little bit and then got a job offer to go over to Google. So I moved to San Francisco. When I got to Google, my day job, and I’ll get that into that in a little bit, was working with small and medium businesses. But when I got to Google, people had heard that I had worked with the United Nations. And so I got drafted onto a project called Billion Acts. And Billion Acts was Google’s initiative to essentially leverage all its resources to try and accelerate social good in the world.
And one of the targets or one of the ambitions for the projects was to inspire a billion acts of service around the world. And so I was part of that team. I can’t take at all full credit or only a small portion of the credit from the team, but in 2015 and 2016, I think some hundred, 200 million people had logged acts of service from the work that we were doing. I got to work with all these Nobel laureates, which was just a very surreal experience for me on a personal level. But yeah, the project got nominated for the Nobel Peace Prize.
which was, I don’t even know, it still feels surreal just bringing up the story again right now. But just an amazing experience during my time there. And then kind of more on the business side, I was working with lot of these small medium businesses on Google’s businesses, which is primarily the big revenue driver at Google was our ads business. And a lot of these small medium businesses struggled with telling their story online, getting photos to kind of showcase their product.
And my boss at the time, this guy named JB, he’s like, Patrick, you’ve done a startup. know, Google isn’t going to solve this problem of helping small, medium businesses tell their story. Why don’t you leave to go start another company and I’ll write you your first check. So he wrote me a hundred K check. We ended up turning it into a $3.1 million seed round. So I left Google. I started my second business called Catalog. We did content production seamlessly for direct to consumer businesses.
Grew the business to over a million dollars in revenue, which for a small business, you that was a good, good early highlight for us. But then COVID hit. And a lot of the clients that we were serving, again, these mom and pop shops, these small businesses, they didn’t survive early COVID. And unfortunately, we were also taken down during that. And I was trying to decide what to do next. And UCLA Law reached out to me. They were starting this new program called the Masters in Legal Studies.
basically law for non-lawyers, you would go in and take classes at the law school. And for me, it was a great opportunity to kind of go back. had a lot of interactions with the law. I did mock trial as a high school student, so always had this passion for law. But yeah, we did that and befriended my now co-founder in the business, this guy named Professor Alex Alben. He teaches AI and internet law. We hit it off.
We co-wrote a paper together on AI and its implications for the US. We started a podcast together. So, you know, I understand the intricacies of doing all of this. And, you know, we asked this basic question of what would it be like to do legal prediction? And that led us to start this business now.
Greg Lambert (13:26)
Nice. So the
moral to this story for all of the high school seniors and college freshmen that are listening is just cold call or cold email some CEOs in some country that you want to go spend the summer in and then just let the opportunities roll in. I don’t know how you did all that and finished your degree at Chicago, but.
Marlene Gebauer (13:39)
in another country that you want to go to and do that. That’s it.
Patrick Ip (13:40)
Hahaha!
Yes,
absolutely. It’s just that easy. Just that easy.
Greg Lambert (13:54)
Sounds like a lot of work, but let’s talk about your work with Alex Alben there when you met him at UCLA. So how did the collaboration on AI and law initiatives lead you to the creation of Theo AI?
Patrick Ip (14:12)
Yeah, you know, I think we were in particular interested around what cases would settle for. And we’re not the first company nowadays to do legal prediction. There’s a company back in 2015 called Legalist, which for the most part has turned into, as I understand it, a litigation finance firm themselves. But there have been a lot of companies that have tried to do legal prediction. And I think on top of AI, obviously advancements in AI, one of the biggest issues has been the area around data.
A lot of the are confidential. If you don’t have that data, it’s really hard to do that prediction exactly. And so I think one of the unique standpoints from the inception of this company is we started it at UCLA law, right? And so we kind of took a research mindset where eventually over the course of the company, we found professors at Stanford law that we started working with. Now we work with Caltech and their AI department. And we basically asked this research question, which is while this data is confidential, a lot of the fact patterns are public. So.
Marlene Gebauer (14:42)
Right? You don’t see it.
Patrick Ip (15:10)
can we create what we in the technical world create called synthetic data? Can we create a data set around what those settlement factors could look like? And so we built that out. We built a prediction engine. And then we started going to litigation finance funds. And they were extremely skeptical. One of the most skeptical bunches that I’ve met, they’re like, we don’t use technology. We don’t trust it. But they gave it a try. They gave us a claim that they had viewed.
that they had funded or not funded. And for the first time, they saw that it was in line with their underwriters. And I think that’s really when the light bulb moment kind of hit like, wow, you know, this team is onto something. And so that was kind of it.
Greg Lambert (15:52)
Did
those litigation finance folks, did they have some numbers that you were able to feed in? Were they giving you information or were you, so, wow. yeah, so, and I know we’ve talked to others about synthetic data and there’s some really good things and potentially some things that may kind of.
Patrick Ip (16:02)
No.
Marlene Gebauer (16:04)
I figure they’re not going to share whatever they have. It’s like they want to keep that to themselves.
Greg Lambert (16:21)
may make the data mediocre with synthetic data. So how were you able to test the synthetic data in order to really figure out, are you in the ballpark?
Marlene Gebauer (16:36)
Yeah, how’d you go about creating it?
Patrick Ip (16:39)
Yeah, I think it’s, so that’s not the only data source that we use, right? So I think if that was the only data source that we use, the direct, the like results may be a little different, but at this point we probably scraped 80 to 90 % of the public results that exist. So, you know, we use that plus what we’re doing with synthetic data. And I think one thing that surprised us in this business building is, you know, as we started to work with the litigation funders and these other clients that we’ve worked with, one thing that
When I was talking to them, I would ask them, oh, do you have a CTO on staff? Nine times out of 10, they would say no. I was like, have you ever looked at any of your past funding decisions or any of your past cases and if they worked out or didn’t, and they would say no. And I was like, what if we were to come in and use our AI talent and AI engineers and actually structure that for training? And so instead of just a gut feel around what cases we should take, you can actually objectively know what works and what didn’t.
And so that’s been kind of the core foundation of our data model nowadays is public data, our research data set, and then the firm level data set that is private to just them. And that is, I would say, allowed us to produce results that has been unseen in the industry so far.
Marlene Gebauer (17:54)
So Theo AI recently announced a 2.2 million pre-seed funding round. So congratulations on that. How do you plan to use all this money to expand? And it’s really not that much money, I know. How do you plan to use this funding to expand Theo’s AI capacities and reach?
Patrick Ip (18:01)
Thank you.
Hahaha!
Yeah, yeah, yeah.
Greg Lambert (18:14)
But we’ll just think what he did with a hundred thousand dollar check from.
Patrick Ip (18:14)
Yeah, I think.
Marlene Gebauer (18:17)
I I’d
Patrick Ip (18:18)
Exactly.
Marlene Gebauer (18:20)
have plenty of reasons to use it, but probably different than yours.
Patrick Ip (18:24)
Yeah,
you know, I think for us is really just trying to figure out how we help the legal industry in general use data, use AI, use predictive analytics to bolster their firms. think when we first started, kind of the early wedge that we worked with were litigation finance firms. And obviously as we mature as a company, when we announced our funding, one thing that changed is we had big law approach us about, how do we leverage this technology? You know, I talked to firms every single day and I think one thing that has started to shift.
in the conversations that I’ve had is people, know, firms will come to me and say, hey, you know, we’re not the most tech savvy, but we know in 2025, we have to use AI to stay ahead. And how can we essentially use this to do something? And I think one of the perks of working at a business of our size, as you kind of mentioned at the top, 2 million is a lot, but it’s also not a lot. But it’s working and collaborating with these new verticals, insurance, big law.
litigation finance, we also do some work in personal injury, where I think what we’ve done extremely well is build the prediction side of the business. But I’ll just give a small example. For litigation funders, you one of the things they tell us is, there’s all these, they get a lot of inbound, but they also want to find great cases. So can we go out and scan dockets proactively? And you know, we’re like, that’s not something we’ve thought about. But yeah, we can, we can do that. It’s basically the same thing.
And so, you know, that’s really where we’re collaborating with clients today to figure out what new extensions of our core business and our core technology to help them do something that they didn’t think was possible before.
Greg Lambert (20:03)
So I wanted to go back. You had mentioned something about the law firms are coming to you saying, we know we have to use AI. Are you finding that they’re coming with enough knowledge about what AI can and can’t do? Because it kind of reminds me of 25 years ago when law firms would go to someone like you and go, Patrick, I need to get on the worldwide web.
I know that that’s, know, I don’t know why, but everyone else is doing it. And so we need to get on there too. Are you finding that they’re, that they’re coming to you more informed now about it, or are you still having to kind of temper or expand expectations?
Patrick Ip (20:31)
Haha
It varies a lot. And I would also say, you know, this business is still new. You know, we started talking to folks last January, and I would say even the course of a year, the mood for the same individual has changed. Right? I think we’re seeing the proliferation of a lot of legal tech and folks are realizing more and more they need to stay ahead of the curve. But I think you also, you know, kind of what we talked about at the top, there’s a lot of fears about using AI. You know, will it replace me?
Is it accurate? Can I trust it? Right. I think there’s a lot of open questions there as well. I think similar to why not kind of what I teed up. You want effective partners in this. You we try to be those for the early clients that we have, which is, we building it responsibly? Do we have the right safeguards in place? Do we have the right training data so that you know, when you look at the predictions, they are actually accurate.
That’s where we try to provide the right level setting for the clients that we work with and alleviate any concerns that they have there. And then I think this is kind of a mutual branch for all of us is what can the technology do? I think one of the hard things with kind ChatGPT o1 and Anthropic is they’ve kind of built a machine where they don’t even know what it’s capable of. And I think that’s the hard part for the entire industry is we don’t know where the upper limits are for this technology.
But I think, you know, I look at Harvey, for example, I think for the first time I got to take a look at their platform and or EvenUp, you know, another darling in the space. One of the things that we’re trying to do along with the other folks in this space is to really make it easy for lawyers to use this technology where they’re not having to write code, they’re not having to be an AI expert, but they can get the benefits and perks ideally in these safe environments to accelerate the work and help them.
Greg Lambert (22:45)
So one of the things that in doing the background check on this is the Theo AI platform aims to compress the case review time, which typically now takes days. And you’re saying you can get that into minutes or even seconds. So do you mind just walking us through a practical example of how the lawyers would use Theo AI to help the legal teams make
better decisions on this or even save time in the workflows.
Patrick Ip (23:20)
Yeah, I would say 80 % of the customers that we work with today are litigation funders. So I’ll maybe start there and then we can abstract to law firms and other abuse cases. But for litigation funders, one of the big issues is they’re having to review thousands of different claims on a daily basis. They don’t know which ones are actually meritorious or ones that they should look at. So the time it takes to read all those complaints, read all the evidence, review the lawyers and law firms that are on there.
Greg Lambert (23:27)
Okay.
Patrick Ip (23:49)
that’s weeks of work, if not more. And so for us, where we can easily ingest that information and say, hey, these are the cases that have a high probability of winning and they meet your investment criteria. And we can do that effectively. Yes. As you kind of mentioned in minutes, that’s huge. And I think what we started to see in law firms, again, we’re still really, really early in the process of understanding how we work with law firms. But of the ones that we started to speak to is let’s say I have a client,
Boeing, and Boeing is facing thousands of pieces of litigation. It takes a lot of work from an associate or a partner to review all those pieces of litigation. And for us, we can scan all of that and say, hey, law firm A, you want to work with Boeing, but this is the case that has high damages where if a great law firm were to step in, maybe you could actually save them all that money. And so that’s some of the early use cases that we’re hearing from law firms.
On top of just using it for client expectations that I kind of mentioned earlier.
Marlene Gebauer (24:52)
I’m curious, like, are you regularly sort of scanning all of those cases to sort of add to your knowledge bank?
Patrick Ip (25:00)
Anything that’s public for the most part, we’re constantly looking at and reviewing. And then also just from the point of testing. Like there was a recent case I just for fun for myself, I think it was an Apple settlement that was done. And I took the complaints and I ran it through our system just to see, I think they settled for 90 million. And I think our system said 80 million was going to be the amount that they were going to settle for. And so, you know, we constantly test and just examine and look like how close or far are we? Are we in the ballpark? Do we need to make adjustments? Like
We’re still, again, in the early innings, but I think the thing that’s exciting for us is the starting point for us, I think, was much higher and more accurate than anyone ever expected.
Greg Lambert (25:40)
I think I know what case that is and I think my phone knows what case that is too. So you talked about the Theo AI is relatively new, what about a year old now or so? So as you’ve been developing it and as you’ve been working with legal professionals, whether they’re litigation funding or law firms, what’s been some kind of support
Marlene Gebauer (25:45)
Yeah, yeah, me too.
Patrick Ip (25:45)
Ha
Mm-hmm. That’s right.
Greg Lambert (26:10)
What’s been most surprising for you in dealing with the market or what kind of insights have they shared with you that have helped you kind of gain knowledge that you probably wouldn’t have otherwise or would have taken you longer?
Patrick Ip (26:24)
I think the thing that strikes me is I think we’ve captured lawyers imagination in the sense, right? Like I think the first thing that I generally hear is like, wow, this is a great idea. Right? Like I think everyone thinks that this product should exist, right? That there should be something where you can input your lawsuit or your claim and get some sort of idea of what other similar cases exist. What is the approximate settlement amount? What are the odds of success? Like
In the early days, we talked a lot about Zillow, like having a, you know, your “Zestimate” on your house. Like, why doesn’t that exist for your legal claims? you know, part of the product inspiration for us is also creating something similar. And so I think that’s the early thing that we hear from a lot of these lawyers in class that we work with. just like, wow, you know, this is really exciting. We want to be part of this future with you to build something great. Outside of that, and yeah, go for it.
Greg Lambert (27:16)
Now, that just, the image
popped in my head of going to a docket and hovering over it and then seeing a little pop-up says, you know, this litigation is worth $90 million.
Patrick Ip (27:29)
Yeah, I think to some degree, yeah, this is a little aside, but this goes to in part like the responsibility we feel as an entrepreneur and like a new entry in the space is also what type of world we’re creating. think, you know, also tees up to your question at the front, which is people are using this potentially as a lawyer. And if there are a lot of legal tech companies, I would say that add litigation to the United States. You know, we’re already a very litigious society, but
Marlene Gebauer (27:30)
is projected to be 90 million, yes.
Patrick Ip (27:58)
They’re automating lawsuits. know, what happens if we live in a world where you’re receiving 10,000 demand letters? You know, is that, you know, is that a world that we want to create? I think for us and you know, why people work at our company, why the investors have backed us is we’re trying to really help only meritorious claims rise to the top, right? Ones that actually make sense to go to court to the negotiated discuss where, you know, for a lot of folks, whether they be on the defense,
or on the plaintiff’s side, they just don’t know if their case is worth being pursued. And so for us, we’re trying to make the system more efficient overall so that everyone can ideally work on something that makes the most sense.
Greg Lambert (28:42)
Well, there’s, and this is completely off the script here, but one of the things that you’ve probably, everyone’s probably noticed about, especially in the United States, is we’ve become much more of a gambling-centric type society. I mean, I went to a Houston Texans football game, and pretty much the entire advertisement on the big scoreboards,
Patrick Ip (29:01)
Yeah.
Greg Lambert (29:11)
were for gambling sites at the game. So it’s like, hey, gamble on the game that you’re watching. And so I could see that a tool like this might play into that where people are looking as like, ooh, this has a 51 % chance of being this. I’m going to be able to take that chance on it. like I said, this was kind of a complete sidebar here on this.
Patrick Ip (29:14)
Yes.
Greg Lambert (29:41)
Are you finding, how are you finding that they’re approaching viewing it? Are they looking at just purely on risk or are they looking at taking chances with that risk?
Patrick Ip (29:53)
I think it depends on who you ask. I’m starting like client-based, like these litigation funders to some degree, right? They’re obviously putting a lot more money into these cases than a general consumer is, but they’re trying to understand their own sense of risk in a completely different way. I think the status quo for a lot of these litigation funders, as I understand it, is you’re looking at backing a case based on the legal merits. You know, is this a case that, you know, does it make sense? But for us, I think we’re giving them a net new insight on data, you know.
how has this case compared to other cases, has it succeeded or not? And I think that’s been really fascinating for these litigation funders because that’s not generally something they’ve reviewed in the past. And so that’s been exciting for them and we’ll see what evolves in the industry from there. And I think the same goes for law firms as well as I again, just want this objective data point. want these in a world where we have so much data analytics, why there should be a product that exists that helps find similar cases
Marlene Gebauer (30:51)
It’s surprising to me that they don’t actually refer to precedent, either their own internal precedent or external precedent in terms of evaluating the likelihood of success and looking at the overall risk. That’s just really surprising to me. I figured that would be just a normal part of the evaluation.
Greg Lambert (31:15)
Do you think it’s because they think the data is between their ears?
Marlene Gebauer (31:20)
Well, that’s a very lawyerly thing, isn’t it? like, you know, I got a gut, seriously, I got a gut sense and like, I know this space and this is what I think. But I mean, they kind of tout themselves as like, we’re looking at data in order to be able to make this evaluation. So that’s just surprising to me.
Patrick Ip (31:30)
Yeah, I think.
I think that’s what we’re trying to empower and augment these firms to do. I think as you kind of said, it’s a no brainer for them to do, maybe being a partial outsider to this space, right? When I look at the composition of the leadership, there’s no one that’s technical. You know, it’s all partners. Everyone’s been a JD. Even some of the AI people that I speak to at these firms, they don’t have computer science backgrounds. They don’t have AI backgrounds for the most part. So I think the legal industry in general is still really early on in trying to understand how to partner with
tech folks like us, would say, or ex-Googlers, right? And that’s where we’re trying to be a proper partner and steward with the legal field is to help them do kind of the things that should have already been done to some degree.
Marlene Gebauer (32:19)
So Patrick, the integration of AI in the legal field comes with challenges. we touched upon this right at the very beginning when we were talking about the generated lawsuit. You were talking about what type of world do we want to live in. So the challenges include ethical challenges, bias, along with data security, and maintaining accountability. So how does Theo AI address these concerns to ensure responsible and effective use of the technology?
Patrick Ip (32:50)
Yeah. So I think when we think about bias, know, and Google got in trouble in terms of its training data set, I think it was late year around bias. I think for us, I think it comes up a little less for our particular company because one of the evolutions of how we transformed as a company is we started really working with the firm level data going back to like, you know, a client that we work with, they’ve, they funded over $200 million in cases.
but they never once look back at which cases were the most successful or had the highest returns or had the lowest returns or which law firms actually worked best or worse for them. And so there is bias. We are creating a bias data set because we want to know what actually worked and what didn’t work. And so I think it’s just about creating the right rule set and the bias set based on your use case. Maybe more of the downside that you don’t want is more of this hallucination.
surfacing the right case when we have a citation? Are we looking at the wrong state or the wrong county or the wrong practice area? And that’s a little bit more difficult to do. I think this goes a little bit more to the evolution of the technology. But I was listening to an interview and the interviewer asked another tech CEO, why does ChatGPT hallucinate?
You know, they should, why can’t GPT just say, I don’t know what the answer is. And I think there’s a bit of a misunderstanding of how the technology works. would equate like the three, five model in the four, a model more to Google auto complete. Right. When you were going in and writing in a sentence, you’ll see that, you know, when you do a Google search, there’s five or six possible next words that it’s trying to predict. That is much more closely tied in terms of a metaphor to how the GPT model. It’s not hallucinating because
It’s trying to answer the question actually how the technology is working on the back end is it’s just trying to auto complete. And so when it predicts the answer, it’s just trying to think what’s the next word that should come in that sentence. Why o1 is really intriguing for all of us in the tech industry is it can go back and think, should I actually be using an auto complete type of machine learning element here? Or should I be using a different way of thinking on the problem? And so, you know, that’s, that’s an evolutionary field. That’s, that’s exciting.
Greg Lambert (35:13)
So I wanted to go back to your time at UCLA because you talked about the Master of Legal Studies.
Patrick Ip (35:19)
a great cut out for me. Is he there for you Marlene?
Marlene Gebauer (35:21)
He cut out for me too.
he was talking a little bit about your Master of Legal Studies from UCLA Law School. And so you obviously have this unique combination of tech and legal expertise. So, and I personally think that’s very important when you’re developing solutions for the legal industry because there are a lot of nuances that if you don’t have that background, you don’t necessarily pick up. So, how has it influenced your approach to develop AI solutions for the legal industry?
Patrick Ip (35:52)
Yeah, I think
one other context point is I went to UCLA law 10 years after I graduated from undergrad. I think a lot of folks, for example, they might go straight into law school after undergrad. I think one of the unique things about the master’s in legal studies for a lot of folks is it catches people in their 30s to 40s. I think the average age of student in my class was 37. And so these are folks that already have a decade plus of experience that are going in with real life.
business ideas and experience with contracts and the law. And the same for me, you I came into the UCLA law program and I think everyone after you do two startups are like, is when is it you’re going to be your third business? And I didn’t know coming out of it, right, that I was going to start a business with my law professor. That was not why I went to UCLA law. But I think when you mix an entrepreneur in this environment where it’s like, why do these things exist? Why is it done this way?
Marlene Gebauer (36:41)
Yes.
Patrick Ip (36:50)
Is there a way to do things better? That’s just a natural, I think, the question set of an entrepreneur. And so, yeah, that’s one of the exciting outcomes of going to law school and then now coming out and working with a lot of lawyers and different folks in the legal profession.
Marlene Gebauer (37:08)
Do you find that there’s a variety of opinions when you’re working with different legal practitioners? Or do they all pretty much are in line?
Patrick Ip (37:20)
Um, I think the one common denominator, I also, because I don’t really come from the industry to some degree, I think there’s some brand perceptions around, uh, personal injury or, you know, different practice areas or different law firms. Like that’s not something I intuitively know or understand. And because I don’t think I come from that, that frame of mind, you know, I generally work with them and like, what is it that you’re trying to do? You know, at your firm, how are you trying to make the process more efficient?
Marlene Gebauer (37:32)
Yeah.
Patrick Ip (37:49)
And I really come in, would say, with a blank slate and say, how can I help make your daily life better? How can I make your client’s life better with using technology? And I’m willing to work with folks from all different types of legal backgrounds to help them be more efficient.
Marlene Gebauer (38:05)
Alright, so I have a fun question for you. So, we understand you’re an amateur triathlete and we obviously know from when we first started that you needed to get your coffee before we started so that you’re also a coffee enthusiast. And I feel these things work in innovation, so I want to know what your thought…
Greg Lambert (38:23)
He’s not just a coffee
enthusiast. think he has actual like a degree or some kind of certification and yeah.
Marlene Gebauer (38:31)
my, okay, well we’re going to hear about this. So how
do we tie this into innovation and entrepreneurship?
Patrick Ip (38:40)
Yeah, maybe I’ll just tie in on the coffee thing. I got when I was in Australia, I enrolled in a coffee college on the weekends and got a degree in express preparation. And actually behind me is a flavor, a coffee flavor wheel, which is if you go to
Marlene Gebauer (38:51)
I was wondering what
that was. It was like a color wheel and I’m like, hmm, what is that? So, okay.
Patrick Ip (38:55)
Yeah. And so
actually my wife is a food scientist. She got her PhD in food science at UC Davis and her lab mate actually made this and I was like, my gosh. You know, that’s so it’s actually signed by her because she was the one that did all the research by this. And so it’s actually now one of my prized possessions because of my coffee background. but yeah, you know, I think I did an Ironman, half Ironman this past year. I think one of the things this is more towards company building.
Marlene Gebauer (39:05)
That’s cool.
I see that.
Patrick Ip (39:24)
But in my second startup, Catalog. I think when you talk to most entrepreneurs or founders, they’re doing their business 24 seven. You know, the folks that I worked with, and when I was down in LA, they were people that I worked with Monday through Friday, but they’re also the people that I hung out with on, you know, Saturday and Sunday. And one of the changes I think post COVID is people, and now I have an 18-month-old baby too. So that’s one of the changes for me as well. Thank you.
Marlene Gebauer (39:52)
Congratulations.
Patrick Ip (39:54)
is I think when people come to work, they want to do their best work at work, but they also want to have a life outside of work, whether it be with their family or with sports or fitness and whatever it may be. I think one of the concepts going into building this third business, and I didn’t really have a term for it, but over the holidays, I was reading about Japanese culture and there’s a term in Japanese culture called Ikigai, which is just generally around finding one’s purpose.
And when we thought about company building here for Theo AI, one of the things that we talk about is I want you to be successful here when you’re working with us at a company, but I also want you to have a full, purposeful, passionate life outside of work too. And how do we support that? I think one of the elements I, as I was reflecting on that was when I worked at Google, one of the things that stands out for, when you meet ex Googlers is they’re not only exceptional at their day job,
but they also have some sort of amazing thing that they’ve done outside of that, whether they were former tennis players or video game champions or elite chess players. And so as I thought about this business too, it’s like, how do I help cultivate that? Because I found for me, to have multiple outlets, have multiple passions, that’s only been a benefit in my career. And so how do I kind of extend that out to this company as well?
Greg Lambert (41:21)
So Patrick, we’re at the point now where we ask all of our guests our crystal ball questions. So I want to kind of pull out your crystal ball, look into the future for us and looking ahead, what do you see are some big challenges or changes, both that you think the industry is going to face or that maybe your work there at Theo AI might face over the next couple of years.
Patrick Ip (41:49)
Yeah, I think on a super high level, when I get asked this question, you know, obviously having a background in AI, think people kind of think of like self-driving cars and flying cars and you know, what are we going to, are we going to have something that’s the equivalent in the legal industry? know, me being a founder and also, you know, talking to other founders in the legal tech space, I think what the future kind of looks like is actually a lot more like the present. Like a lot of the, I look at Spellbook, for example, a lot of their innovations are in Microsoft Word.
And a lot of the AI development is in Microsoft Word. When we started out, we built out a platform and similar to Spellbook, we found out that people wanted the insights and the predictions in email. And so we started building email integrations where people didn’t have to log into another platform to get their prediction reports. And so a lot of the new AI powered workflows as it is pertains to the legal side.
is just going to be in where you already are working, whether it be Microsoft Outlook or Microsoft Teams. And so a lot of the AI advancements are going to, they’re going to look like what you’re already used to. They’re just going to be a little bit smarter, I think is where we’re going to end up.
Greg Lambert (43:00)
Well Patrick Ip from Theo AI, I want to thank you very much for coming in and talking with us. This has been a lot of fun.
Marlene Gebauer (43:08)
Mm-hmm.
Patrick Ip (43:09)
Absolutely. Thanks for having me.
Marlene Gebauer (43:11)
Thanks, Patrick. And of course, thanks to all of you, our listeners, for taking the time to listen to the Geek in Review podcast. If you enjoy the show, share it with a colleague. We love to hear from you, so reach out to us on LinkedIn.
Greg Lambert (43:23)
And
Patrick, we’ll put links in the show notes so that people can find you that way. But what’s the best way for people to find out more about Theo AI or to connect with you directly?
Patrick Ip (43:34)
Yeah, through our website, THEOAI.AI that’s going to be the easiest. I will say one of the secrets of that, considering our stage, is actually if you submit a support ticket or a contact, it actually goes straight into my inbox. So that’s an easy way to get a hold of me. But yeah.
Marlene Gebauer (43:48)
You shouldn’t have told everybody that.
Greg Lambert (43:51)
We know the secret sauce now.
Marlene Gebauer (43:56)
well, thank you again Patrick and as always the music you hear is from Jerry David DeCicca. Thank you Jerry.
Greg Lambert (44:02)
Thanks Jerry.
Alright Marlene, I’ll talk to you later.
Marlene Gebauer (44:04)
Okay, bye.
Patrick Ip (44:05)
Thank you.
