Building Autonomous AI for Legal Contract Review

Shantanu Singh: The user experience that we hear is it's magic. Like I, I can't believe it did all of this in however many minutes. So that's fun to hear.

Jacob Miller: Hey everyone, and welcome back to the Startup Wisconsin Podcast, a show where you can learn about Wisconsin's growing tech scene through stories of startups, founders, investors, and the talented people making it all happen.

Today's conversation is with Shantanu Singh, a co-founder of Obviate.ai. After years of practicing law in Wisconsin and Illinois, Shantanu saw firsthand how even sophisticated legal professionals were drowning in mechanical contract work. When OpenAI released Chat GPT, he described it as discovering fire.

Suddenly, automation of his own job became possible, but obviate isn't just another AI wrapper, Shantanu and his technical co-founder Tom, filed a provisional patent on their approach, which involves multiple APIs. LLM. Judges and autonomous agents working behind the scenes in their philosophy create an easy button that requires no legal education, no prompt engineering, and works outside any single ecosystem.

It's designed for small businesses and solo lawyers who don't have any kind of massive legal department backing them up. We talk about the misconceptions around AI and legal work, why most AI products are automated rather than truly autonomous, and what makes Milwaukee's tight-knit AI founder community so valuable.

Alright folks, let's dive in.

Shantanu Singh: I am a startup, uh, co-founder of Obviate.ai. It's a legal tech tool that was something I hope I had when I was a practicing law. It, uh, reads, analyzes, and suggests. Risks to mitigate and gives you the edits as well for your consideration. All without prompting and no typing. Uh, it's built for small businesses, solo lawyers and law firms, and it really excels at transforming dense agreements into clear, actionable insights that you can then, uh, leverage to complete tasks that usually take hours and two minutes.

And our team is on a mission to really make powerful legal tech accessible to. The people doing the work, not just with the law departments that are standing behind those who can afford them.

Jacob Miller: You're a licensed attorney, but now you're leading or helping grow this AI startup. Mm-hmm. I kind of want you to walk us through that journey, you know, when did the idea for obviate kind of first start to take shape and talk about that a little bit.

Shantanu Singh: Before I was a lawyer, I was a scientist, so I went to Washington University in St. Louis where I studied, uh, cell biology. So I was doing clinical research on Crohn's disease and ulcerative colitis. So I've always had a passion for discovery and working on, uh, I guess hard science problems. And so the legal training is very good for, um, identifying how to navigate an ambiguity.

Uh, the science training is very good for how to scaffold some logic and some repeatable systems. To make your experiences with ambiguity fewer and fewer 'cause you're just continuously scaffolding ways to address it more and more. Mm-hmm. Um, and the practice of law is very suitable for, um, uh, like deductive logic.

Sure. Uh, it's, it's suitable for scientific thinking actually, but it is used for, uh, navigating the gray. That's why people hire lawyers. They don't hire lawyers for resolution of, you know, yes, no questions. They hire lawyers because the yes no answers depend upon what side you're on. And so you have to hire them for persuasion.

When I worked at a company that was in the financial services space, they had several hackathons internally at the company. And so I thought that was cool. So I had previously worked at a, um, software company before I went to this financial services company in Wisconsin. And at this hackathon we built a tool, this is around 2014, we built a tool that could identify risks and open source licensing, um, for individuals who wanted to know if they had specific obligations relative to certain open source licenses.

Mm-hmm. And so that got me into natural language processing, natural language understanding, um, like linear algebra and. Tokenization. And so that slowly developed, uh, into other side projects. And that resulted ultimately about three years ago when OpenAI released its GPT. Mm-hmm. It, it was, for me it was like discovering fire.

Jacob Miller: Yeah,

Shantanu Singh: for sure. And, um, because I immediately saw that this could automate my job and it wasn't something that I was running away from, I was like running to it. Mm-hmm. 'cause lawyers handle some really complicated issues, but they also handle a lot of mechanical issues. And so there's a lot of software you can build, I believe in the legal tech space to enable high level thinking and um.

Offload the, the mechanical work. And I, it's not mechanical in the sense that it's administrative, but it, it's just repetitive, but it takes a lot of

Jacob Miller: time. Yeah. Can you share an example of that just for anyone listening that might not understand what that means?

Shantanu Singh: Classic example is in litigation. There's this,

in litigation, there's this concept of, uh, discovery. Um, I'm sure people have seen movies where they see lawyers going into these warehouses. There's documents everywhere.

Mm-hmm.

There's hundreds of 'em. That's actually not that off base, but that was like 20, 25 years ago. That's rote work. Like someone has to do it.

Jacob Miller: Yeah.

Shantanu Singh: About 15 years ago, that process was enabled with machine learning. And so you are in some platforms you can actually discover, um. Relationship between documents, even if physically they would've been separated by, in my case and one project they had like hundreds of miles. Yeah. Um, so it's saving drive time, it's saving on site time, but most of all, it's getting you what the client it's getting for you, what the client needs to bring their case.

Jacob Miller: Yeah.

Shantanu Singh: So that's an example of like necessary work that has to be done. But if you can find those documents or whatever evidence you're looking for more quickly, that means you have a better, you have more time to build a stronger case through analysis.

Jacob Miller: Sure. Was there like a specific moment, uh. When you were like, there has to be a better way to do this.

Was there, I dunno if there was a case that you were working on or just there was, you know, obviously you talked about, um, some of those hackathons inside the, those organizations that you were working at as well. Was it, was there like a specific moment, like aha moment or was it just kind of like over time it just built up to like, okay, this totally makes sense for me and what, what I wanna do next?

Shantanu Singh: No one person can be good at everything. Mm-hmm. Right. But you focus on what you can be good at and then you really try to amplify it. And so what I am good at is I can take disparate issues and build technology around them. I don't, it's just something that I do. And so this was a natural, um, this project, if I had to say, when did it first like, come to my mind, um.

It actually first came to my mind when I saw other solutions like eight years ago, trying to do this type of legal analysis and contract review. And if you studied machine learning and, and deep learning, you would, you would find out that these machines are not as good as lawyers.

Hmm.

Because you're essentially taking your internal workflow, which is cognitive and inside your head.

Yeah. And now you're projecting all these switching costs by interacting with a digitally user interface to do this same thing that you are just naturally doing in, in a less modern way, which is typing and looking at a screen and taking notes on the margins of, uh, paper. So that was when I first was like, okay, this is possible.

But I don't think that technology exists yet. Um, so I started then, uh, focusing on like workflow automation. Hmm. And how do you distribute work within a large law department? Hmm. How do you distribute work? Um, in a, in a, in a different setting? Um, so I'd say it was seeing other projects and seeing that they're limited.

Then, um, seeing that you gotta meet the technology where it is and make the best of it. So that to me was workflow automation. Um, and then once chat sheet PT came out, um, you know, I feel fortunate that I had like that, these insights 'cause they, I wish I could say it was like a culmination of things. Sure.

Um, but it wasn't, it was just a. Uh, and this is what I tell folks that I mentor, is like, your job is just to drop as many dots as you can in your daily life. Mm-hmm. And they'll, and they'll connect. You just don't know how.

Jacob Miller: Yeah.

Shantanu Singh: But you just, you need to drop all the, all the dots you can possibly drop. And that means getting outside of your comfort zone, trying something new.

And it's not about succeeding or failing, it's the fact that you're developing this connective tissue that one day is going to be strong enough for you to do something. Yeah. And, and so don't limit yourself to stuff like, I need to do wax and that's it. No. It's like, I need to do A, B, C, D, E, F, G, all the way to Z and then go again.

Yeah,

Jacob Miller: yeah, yeah.

Shantanu Singh: And, and see what's happening. Um, yeah. That's my philosophy.

Jacob Miller: Yeah. I would agree. I, I feel like I feel very similar. Um, my background was in like video production, photography, and audio. Mm-hmm. And then I kind of like worked my way into like marketing management strategy and stuff like that through the work I was doing.

'cause I just always saw like the bigger picture, I saw the potential for new connections, for just ideas and how they can be executed. Mm-hmm. Like collaboration with other people, other organizations, other skills, like things like that. I just, my brain just started to see patterns of, of like, oh, here's like a different way we could do it.

And it could be even better than, you know, we originally anticipated kind of thing. But the only way you come up with those, like mini rev revelations is kind of by experimenting in all these different kind of like verticals or fields or whatever. And if I wouldn't to have dabbled in all these different things or met all these different people within my kind of career arc mm-hmm.

I wouldn't be here and I wouldn't be talking to someone like you, you know? So it's like, it's just, I find it so fascinating. I, I totally agree. So, yeah. Um, so I am, I kind of wanna go back into like the product itself. Mm-hmm. Um, from what I can see, like on the website or just what you've shared with me. Um, it sounds like you are using, um.

AI agents, so autonomous agents that are kind of like working behind the scenes doing some actions, because you said it doesn't require prompting, or maybe there is some, but it doesn't seem like there's a lot like you would GPT or clot or something like that. Um, but what, you know, what I, what, what is kind of happening there?

Like when, you know, if someone like buys your product and they're using it, kind of what's kind of going on behind the scenes there?

Shantanu Singh: Yeah. Um, so the technology we built, we filed a provisional patent on, and at a high level, we abstract as much from the user as needed to make this function in a truly autonomous way.

Okay. And so how do we do that? There are prompts, but those are in the logic of the software and we don't just use one agent. We use, or excuse me, not one agent. We don't use one API, we use multiple APIs and there's concepts of LLM as judges. There's uh, concepts of chain of thought, tree of thought, all of that stuff.

I, without getting into specifics, we do like pretty, um, robust, uh, prompt design. And my co-founder and I, Tom, you know, have discussions of like, we, we started designing this thing, um, back in like October of 2023. And the genesis of our idea was a paper by, uh, the owners and publishers of Minecraft and Nvidia.

Um, the paper's called, uh, Voyager, and so they've made it a multimedia pla, um, paper now. Mm-hmm. So they actually te they actually show you essentially what, what is the gist of the game? It's, it's basically a minor who figures out how to build his or her world without any human intervention.

Mm-hmm.

And it, it, the paper walks you through this concept of sensors and this concept of reinforcement learning, um, and how a couple inputs can scale the ability to build a universe.

Right. And so that's what we tried to do, somewhat like the Staples button from those commercials back in the day, is we literally wanted to make an easy button.

Jacob Miller: Yeah.

Shantanu Singh: And,

and like that was our philosophy. So. How it works is you upload a contract and our application is not a wrapper. So a lot of folks talk about AI wrappers.

Mm-hmm. We

are definitely not that. Um, how come, how can you feel positive that you're not that Well, and if you're a true AI wrapper, you're just a presentation layer when you put an app out there. Yeah. Like, there's not much going on except you're creating a GUI to like hit a button and that executes a prompt that is programmatically working through an API.

But you have to do it iteratively. Like you have to go one by one by one. Um, so we, we wanted to build something that a, um, required no, like legal education. Right. But it, it's also meant for, it's also meant for people who have legal education. Um, two, um, that. You don't have to know how to design a prompt because that itself is a profession now.

Um, so we wanna do all that. And three is we did not want the user to have to do track changes in Microsoft Word manually.

Jacob Miller: Yeah. Yeah.

Shantanu Singh: We wanted that to be done for them, but we also wanted it to be done outside of the Microsoft Office system. And so what we, and why did we do that? If you go into any sort of environment or ecosystem, whether it's Microsoft, Linux, wherever it may be, um, Macintosh even, um, what you'll see is, uh, that we are, how should I say this?

What, what you'll see is folks are trying to.

When they go into an ecosystem, the flexibility of the apps that they wanna use are curtailed, they're limited. Um, so if you want to be a plugin, let's say Microsoft Word, you have to make a trade off between meeting people where they work versus meeting people, um, with what they actually wanna do. Right.

And so for me, that was, those are the four things. Like they don't have to have a legal education. It's a self, it can be used for self-help. Um, two, they don't have to prompt, they don't have to type, and it can't be inside any ecosystem. It, it has to generate a data object that can be interoperable within any other platform.

Um. So those were the four guiding principles. And um, literally using that paper as a guide, it's, it's not really a recipe. It's like we took it more as motivation, um, that it's possible. Mm-hmm. And so that's where we went. And, um, we, like, we make, I mean there's, there's hundreds of API calls, uh, and the second thing that I think you asked was, well, how's it work?

So I was telling you more of like, like what was our motivation and now how does it work? Mm-hmm.

Um,

we are, I have a strong opinion, as does my co-founder, um, that what a lot of people are building is not agentic, it is automated. And so what does that mean? If, if you Google AI automated versus AI autonomous, you're not going to get a single standard.

Uh, one's focused on the technology. The second is focused on the outcome. Hmm. Okay. And so what is the outcome that you want? Well, if you do what everybody's doing, in my opinion, it's changing now because the builders of ap, the builders of Grok and Gemini and all those, they're actually building agentic into their actual chatbot.

Um, but most, most applications just do one thing with multiple steps very well. Right? So they give you the recipe and, but you have to bring the ingredients. And, and what are those ingredients? What's your prompts? It's how do you, how do you wanna design your particular solution? In this constrained environment of an automated AI platform.

So what's autonomous? Well, autonomous. Is it how much of what the AI does can be done without any human intervention, right? Mm-hmm. Um, so if you take a, if you take, you know, electric cars and they're saying they're autonomous, well, everyone knows that, like true autonomous automobiles are ones where you don't have to do anything in the automobile.

But automated vehicles are those where you do have to give it something here and there, like taking the steering wheel, um, or wherever it may be. Mm-hmm. And, and so with, with knowledge work, our strong opinion is knowledge work cannot be truly, truly autonomous, like a hundred percent. And the reason is.

Knowledge workers are paid for their knowledge. And so there has to be a component of explainability when they do their work without ai. So if someone delivers something, in my case, if I deliver a memo or a legal brief, I have to explain why I took certain steps and not take certain steps. So there's an explainability component and I think if you operate within that constraint, I need to be autonomous, but I need to have explainability.

Mm-hmm. Right. Those two constraints I think really help product developers and product designers make some really cool stuff. 'cause I have seen some really cool stuff that,

Jacob Miller: yeah,

Shantanu Singh: that does hit that point. Like it's autonomous, but it, there's also a lot of explainability.

Jacob Miller: Yeah. What do you feel like is kind of the biggest misconception around people using AI for legal tasks?

Um, I, I would imagine it's very similar to something within like healthcare with HIPAA compliance and stuff like that. Is there, do you feel like, 'cause obviously you're not building with a wrapper, so there's that. Mm-hmm. Um, what are your thoughts there with any misconceptions people have currently, like as you've been having conversations with customers, maybe as you're trying to build trust with them, what are they saying?

Like, oh actually, like we can't just do that 'cause of X, Y, Z and then you kind of say, actually this is how we approach it and this is why it actually works.

Shantanu Singh: I used to watch college football religiously when I lived in Illinois and um. Jim Thorpe, there's a real famous photo of him, uh, with like these dingy shoes on, and I don't even think they're matching.

Uh, and he won, I think it was the gold medal, um, in those shoes. So when you're, when you're mentioning like what is, what is the misconception, right? The misconception is that it is a user problem, not an AI problem. Hmm. So I kind of mentioned Thorpe and the reverse, like it didn't matter what tool he had, it was him and his endurance and his determination that led him to be a great athlete.

Um, the same thing with ai. You can have the greatest AI tool and not take advantage of it at all, or you can take advantage of it and not do really good things because you didn't know the constraints. Hmm. Um, so for me, the legal problems are that, I shouldn't say they're legal problems, legal professionals who use AI and don't understand the tool they're using.

They're the ones that you read about in the press. Hmm. Um, I mean, you could ask nine out of 10 lawyers, I hope, 10 out of 10 lawyers in the courtroom. Um, would you ever submit anything to a judge without reading it? I mean, 10 of most people are you, most people say, no, of course not. And so these habits that people have of, I'm just gonna submit stuff that my associate wrote, um, that's, that's what gets him in trouble.

So to me, it's knowing the constraints of the tool. Knowing your personal habits as a professional, whether you're using AI or not. Sure. And then seeing the risks from those two points. What's the constraint and what, do I have a habit that's going to magnify, uh, those constraints and create a problem for me?

Jacob Miller: Yeah, I mean, how are you kind of right now building trust with clients with this kind of new tool? Is it, I mean, are you, do you have case studies? Are you, like, is it just once they kind of see it in action, they're like, this is awesome. I'm, I'm kind of curious what that has been for you right now as you're kind of, uh, going to market with it and who, who seems to be more attracted to it as far as, uh, different customer verticals that you're going after?

I'm kind of curious how that's going.

Shantanu Singh: Yeah. Um, so what we have are a lot of testimonials. Which we're looking forward to publishing on our marketing page that focus on, um, like two points. One is the user experience, and then two is the substance. The user experience that we hear is like, it's magic. Like, I, I can't believe it did all of this in mm-hmm.

However many minutes. And so that's fun to hear. Um, what do people get out of it? This is a common fact pattern, which is also a issue for us retaining customers, is we as a beta release, are providing our solution at no cost. For, for a period of time. Okay. And so what we started seeing, um, was a couple habits.

One was people would, users would just blow through all of their contracts over a given two day period. So you, so you would have like five or six agreements that were, uh, you know, just negotiated through our platform. Yeah. So that showed us two things. One, that our assumption of is it reliable, was playing true for certain verticals.

And when it came to small business owners, it did ring true because they were throwing in contracts that were already completed, but they were throwing them in there to quality control. The work of their prior counsel or their own executive team that negotiated those deals. And so they would get, it would be a sanity check.

They would see, hey, like we, we kind of missed these issues, but luckily we haven't approached to resolve 'em. And then there were folks who wanted to just get their contracts through for closing deals. And they had a lot of fun with it. They got a lot of value with it. They gave us a lot of testimonials.

But the problem was, and this is what a lot of AI products are encountering, is the AI works so fast that you can accomplish so much in so little time that your users may not need you for a subscription fee. Right? Yeah. You, you, you might, you might have to go to unit pricing depending upon what vertical you're in.

So the three verticals that we're in. Predominantly are, uh, small to medium sized business owners, uh, solo lawyers, and, uh. In-house lawyers, okay? All three of them have different, um, they have different ways of approaching the same issue of contract negotiation.

Jacob Miller: Today's episode is brought to you by headway, a digital product agency based in Wisconsin.

Do you need to design and build a world-class user experience for your software, but feeling like you just can't get there on your own? That's where headway comes in. We're the folks who help ambitious startups and enterprise teams bring their product ideas to life through design, development and product strategy.

We don't just give you user interface designs and leave you to figure out the rest. We work with you and your target customers to create beautiful user experiences, build scalable design systems, and provide Silicon Valley level talent to get it done right the first time. Whether you're launching a new software or just looking to add features or we've got your back.

We've worked with startups in industries like logistics, healthcare, FinTech, and EdTech, helping them solve their biggest software challenges. And here's the thing, no matter how complex your product is, we can help you gain clarity and get the job done together. We're so confident that we'll be the right fit, that we offer a 50% money back guarantee.

If you don't think we're working out within the first 45 days, ready to see how we can help you and your team, just head over to headway.io and book a free consultation Today. You were sharing, you're the, the, the founder with, uh, industry expertise mm-hmm. Uh, as a lawyer. And obviously you talked about, um, some other things in your background, but also you have a technical co-founder.

Mm-hmm. Um, maybe just talk about that relationship a little bit. Yeah. And how you guys kind of manage or approach like shared responsibilities and um, kind of like how you make decisions together. 'cause obviously there's that, that dynamic where you have to say, Hey, we gotta make a hard decision right now.

How do you kind of handle that or just kind of curious?

Shantanu Singh: You can tell by my beard. I'm not a young guy anymore, but, um, so my, my co-founder and I have both worked as professionals in whatever industry. We're in myself for 25 years, I think Tom, who's my co-founder for, you know, more than 35. And we worked together and we still do actually in our day jobs, um, at the same place that we met, which was about 10 years ago.

And he and I have participated in some of those hackathons I mentioned and Oh, cool. Okay. And we built some really cool stuff and so we both have like a passion for AI before it became cool. Uh mm-hmm. Like before, like when you actually had to read all of those books by O'Reilly and Manning and

Yeah, yeah, yeah.

Uh, and like get deep into how these things work, or at least I did. Um, and he and I as deep professionals, we have, um, something that I think younger teams may not have, which is, uh, we exercise discretion as best as we can, uh mm-hmm. With informed judgment. And so I'm not a big fan of move fast and break things because some things that you break have a margin of safety that's so thin that it's game over.

Mm.

Yeah. Um, and so you have to be intelligent about how you iterate and experiment. And so the concept of move fast and break things. It's a very nuanced like phrase, but at the end of the day, it's not a one size fits all approach to product development. Yeah. So why am I, why am I focused on product development?

Because we're solving a very difficult problem with deep expertise in the law and in ai

and yeah.

So my technical co-founder, he's built, managed, governed, architected technology systems that have a material, have a significant impact on many people's lives. And

yeah.

Uh, so we, we bring a sense of like discipline and maturity and we are our own like worst critics.

I bet A lot of our ideas are on the floor, uh, particularly mine, like, which is why we get along so well, is he's very. Task driven, mission oriented, and yeah. Uh, I'm more like, you know, thinking about stuff that we can do. He brings me back to reality.

Jacob Miller: Sure. Yeah. Yeah. I get it. I'm the same way. So I always have more ideas.

That's, that's my problem.

Shantanu Singh: Yeah. Ideas, after a while, you begin to realize are,

they really don't mean anything, um, unless you act on them. That's right. And any, if you have ideas and you don't act on them, that's fine. It just means it might not be the right idea for you. But if you, mm-hmm. But if you have the gift of coming up with a lot of ideas, then I think you have a responsibility to really.

Have your own system of evaluating whether that idea is actionable, viable, whether you have the capability or time to do it. Mm-hmm.

Jacob Miller: Yeah. We, um, where I work, uh, it's called headway. We have a, uh, philosophy called craft within context. And that's understanding, you know, your, your time constraints, your budget constraints, your skill constraints, like all those things Okay.

With, with what we have, what is possible, what can we actually follow through on and promise to be real right? Uh, and I just think that's, that's such a, and I've tried to do that within the work that I'm doing, whether it's stir up Wisconsin with headway or anything in life. I'm like, okay, well here's what we got.

How do we go from A to B with what we got? Mm-hmm. So

Shantanu Singh: yeah, I think that's gonna be like a future skillset is, um, even though AI will be able to do a lot of things, uh. There was like a famous Hollywood actor, um, who was asked like, is AI gonna take over Hollywood? And he mentioned, you know, there is a layer that's gonna get really disrupted.

Jacob Miller: Oh yeah, for sure. But there

Shantanu Singh: isn't going to be a wholesale replacement of mm-hmm. Directors, actors, and filmmakers. Because there's one thing AI can't do, which is have good taste. Mm-hmm. It needs to, it doesn't have the ability to be situ situationally aware of good taste.

Yeah.

Right? Yep. But humans do. And but only if they're trained at it, if they study it, if they are exposed to it.

And I think that's like the future skillset of this new AI world is humans that can develop and project good taste as a user experience. Humanize what they're trying to get done with AI will be the winners.

Jacob Miller: Yeah, yeah. I, I'm a very firm believer in, there's also, you know, there are always, like, for example, you were saying like some parts of the film industry get just replaced kinda like a lower, lower end of the spectrum, so to speak, um, with the output of films.

Uh, and people will watch that stuff 'cause people always watch garbage films because it just exists and it's there. But I do think that humanity always wants to support humanity. And so when it comes to creativity and creative output, whether it's music, film, obviously like theater performance, you can't just like replace that with technology.

Like that's like a human body on a stage doing something incredible like that's gonna be there. So I, I do feel like. There's more cool technology that, and there's gonna be cool new forms of entertainment, which is awesome. But I still think that, like you said, there's gonna be, the human craft is gonna be very well supported, but also those directors and those creators are gonna like, be able to do even more, but also apply their taste to these new things, which is cool.

Mm-hmm. Um, uh, I actually wanna talk about, uh, what's been harder than expected for you and your co-founder as you've been building this product? I, I actually don't know if you were both full-time on this, if you're part-time. I'm kind of curious. Um, what's been difficult, you know, the past, maybe we'll just talk about maybe the past year or six months or so with the product,

Shantanu Singh: having two jobs and only getting paid for one.

Yeah. Yep. Um, so a lot of like sweat equity, a lot of. The day doesn't stop at six when I come home. Thankfully, I have a very understanding and supportive wife and kids. And so I, I crank out mo, I crank out a lot of work at night and in the early morning. Um, you know, your body tells you when you need more sleep, but I think most founders are all sleep deprived.

Jacob Miller: I'm sure. Yeah.

Shantanu Singh: And you just, and you just accept it at certain point and then your body tells you, okay, I think I'm about to crash. Uh, that's number one. Number two, so the time issue number two is the, the ecosystem of startups in Wisconsin, um, is growing faster than. What the institutional pipelines are developing it.

So what do I mean by that? So like from engineer to founder, that pipeline is getting better in Wisconsin. It's obviously not the same as it is in Texas, New York, Illinois, or Washington State, California. Um, so I, I find it hard that there isn't that much connective tissue to like, support startups that are going to make startups like really, really scale.

Mm-hmm. And I don't think it's an issue of there's not enough good ideas. I don't think that's it at all. Um, if you walk into any lecture hall at the University of Wisconsin system, Marquette, wherever you go, and you just spend a week there. I, I will bet money that a creative founder will find an, I will find a product to build.

Yeah. Just by

sitting in classes. And so to me it's more like an infrastructure issue of how do you support, like dreamers like that, how do you support entrepreneurs like that? Mm-hmm. The access to capital is one thing. Um, sure. The, the second thing is meeting folks who are going to learn with you. 'cause everybody's learning are going to learn with you how to make good bets, right?

Mm-hmm. And I think mm-hmm. Those other ecosystems that I mentioned are not successful just because they have VCs and they're liquid and they have. Extremely, um, wealthy exits and the founders staying those communities and then they fund other startups. Mm-hmm. That cycle definitely does exist out there, but it also is important to know how to make a good bet.

Right? Like, what constitutes a risk, value, um, proposition that's worth it for an angel investor or that's worth it for a group of investor. So I think that's something that was kind of, uh, shocking to me is, um, I shouldn't say shocking. It, it, it was out of pattern. I didn't expect it. Um, and the other item is a lot of folks want to like, um, help, but the definition of help is something that I had to learn.

So I could ask them. Clearly not. So they felt uncomfortable, but that they felt empowered to help. Mm-hmm. Right. And I think a lot of that was just learning by making mistakes. So money is like, not, you can always find money, in my opinion. Um, it's, it's 'cause the, there's a global VC system. Uh, but there is a need for an ecosystem because I do believe in the Wisconsin idea that they taught at, um, UW medicine, right?

Went to law school. Um, which is like, can you do something that's within the state that's good for this state? Like state and take it and take the resources that the state offers and build something with it for it. And that's aligned with the state's history if possible. Um. So for me it's more of like, I wish we could work together as an ecosystem, whether it's in Milwaukee, to Green Bay, Madison, Appleton, Oshkosh, wherever it may be.

Yeah. Um, to develop an ecosystem where people are learning how to make good bets.

Jacob Miller: Yeah. What do you think it would take to make that like, more real? I guess

Shantanu Singh: Like stuff like what you're doing. Stuff like what I'm doing. Like, I don't know you, you don't know me. Um, but we're all pursuing the same like sunlight.

Uh, we're all, we're all like trying to strive for something similar.

Yeah.

It's taking a bet like that. Like, there are probably some podcasters that I should not be on. Um, yeah, I get that. Yeah. And, uh, so how do you learn how to make, how to make that bet of exposure? Right. Um, so every day you're making, you know, a risk analysis of is this worth it?

And having a community that's focused on entrepreneurship and venture capital backed companies that are not totally premised on legacies and logos, which this community focuses, um, on, not, I think by intention j just by history and inertia. Um, sure is, is what needs to happen. Like people should not be afraid to make good bets, but they need to learn how.

Jacob Miller: Yeah. Yeah. I mean, we've been talking, uh, at Startup Wisconsin and with, with other folks, you know, in Milwaukee and within Madison and about how, 'cause the, the, the, I've had this conversation before about like, there's a ton of resources here. There's a ton of people here that want you to succeed, that capital's either here or elsewhere, right?

Um, but like, how do we, when someone like raises their hand and says, I'm, I wanna be a founder, I have an idea, I wanna pitch my idea or whatever, um, it's really hard to get them to the resources or people that they need faster. Um, so something that we've been kind of, uh, uh, pre-planning and trying to figure out how to make it work is almost like a concierge role at Startup of Wisconsin where, you know, startup of Wisconsin's, like the front door to the ecosystem.

Then we get to know you, you know, and maybe we use some kind of AI tool to help record these conversations and help connect dots faster, you know, so to speak. Um, and say, Hey, now that we know a little bit about you, here's three people we're gonna introduce you to based on your current needs right now and maybe future needs in the next three to six months.

And then here's an accelerator program. We recommend you check it out right now, like immediately deadlines in two weeks or what, you know, whatever the scenario is and the timeline. But it's like, the idea is like, how do we increase their chance of luck by getting them connected to the right people and the right resources sooner than later instead of them fumbling through Google, fumbling through GPT, trying to understand what they should be doing, how they should be doing it, um, finding a, a mentor for them.

Like all those things. Um, again, we can't guarantee success. We can increase luck and like that's, that's all we can really aim for. 'cause you know, as much as we can say, Hey, if we do this program, this is the economic impact, we'll we'll make, we know ultimately it's up to that founder if they're gonna succeed or not.

Right? Yeah. We can't guarantee that. Like, just because we give you everything you need. Oh, you got funding and you got an through the accelerator and whatever. Like, it doesn't mean anything until they actually went, you know, so, so yeah, that's kind of something we've been thinking through and it, it was, it's always, I dunno if it's the phrase is nice to hear, but it's, it's been, uh, a repeating kind of conversation we've been having.

Like, I wish there was a way to like connect these dots faster. Um, so yeah, it's just good to hear, I guess. I don't know.

Shantanu Singh: Yeah, I mean, I'm happy to help that effort. Um, and to any of your listeners, you know, I'm happy to, if they're having issues with their contracts, um. Check out our tool. It's,

Jacob Miller: yeah, absolutely.

Shantanu Singh: You know, I think that, um, how should I, if again, this is another movie, um, analogy. Um, you know, there was one director, he's, he's kind of famous for independent films and, uh, he, uh, had this comment that Don't wait for a tribe to rescue you. Build your own tribe. And like, so for him it was like, make movies with your friends.

Grow up with your friends. Like they, if the, find the people who like really passionate about movies and filmmaking and make bad movies, but you'll grow and you'll develop a, a method of like what it is to be a filmmaker. And the same thing in entrepreneurship. It's like, build something with your.

Ecosystem that you know, that you trust, you'll make mistakes. But there need, and I mentioned this phrase earlier, the margin of safety, like the wider margin of safety entrepreneurs have in any given ecosystem, the wider the margin of error is. Mm-hmm. Right? And people confuse those concepts. Like what is a margin of safety and what is a margin of error?

Margin of safety is like, I can really screw this up and nothing like this. This is not an irreversible decision. Right?

Yeah.

So your margin of error to make bad bets is pretty high.

Mm-hmm.

Uh, but if your margin of safety is like we, I, if we make a mistake, it's not relevant, whether it's reversible or irreversible, we have limited time and we can't lose that time.

And if we do, it's over. Yeah. Right. So that margin of error to making a mistake is like super tight, you know? And I think if we focus in any community, and particularly here in Wisconsin, like how do you create a margin of safety that's wide? So when dots do connect, there is a place where people can experiment.

And if they do pursue an idea and it doesn't work out whether a business model or technology or both mm-hmm. It's not the end of the day. It's not a data point to say no, it's a data point to say how do we improve this? Like mm-hmm. How do we de-risk these opportunities for folks who are making bets on themselves?

Jacob Miller: How has it been for you, you know, building this in Milwaukee? Um, I believe that's where you're building it from. Right. And is your co-founder also in Milwaukee too? Mm-hmm. Yeah. So what, what has it been, how has it been for you in Milwaukee? Um, do you go to like, you know, any of the local events that are there?

Obviously you talked about the hackathons that were within your organization. Um, but I kind of, I'm kind of curious to your thoughts and feelings around any of those if you've attended anything.

Shantanu Singh: I am going to them more frequently or getting invited to more national and, you know, other conferences. Uh, but that, that's a lot of time.

Mm-hmm. And that's a lot of money. And for me it makes more sense because we do have two jobs, is I have to interact with customers, which we do, and prospects to learn what is not working and what is working. Mm-hmm. And I have to just go out in the world and meet those people. Um, but whenever someone asks for help, I'm more than willing to explain, you know, how we built this.

Uh,

yeah.

As much as I can bounce ideas off, like there's, there's a local group of AI co-founders in Milwaukee. Um, I know most of them. And yeah, we bounce ideas off of each other. We walk through our prototypes, we tell each other how it can improve. We explain what's an existential threat to the business model that's cool, or to the technology.

So it's, it's not like wide for, for me, it's more focused and precise.

Jacob Miller: Yeah. That, that group that you're talking about, are you guys just engaging through like Slack or through, I'm just kind of curious how you're exchanging those things, sharing feedback?

Shantanu Singh: Both. Both. Both. I've, I, I know these folks and we meet in person coffee shops.

Oh, cool. Well, sometimes Is this like a

Jacob Miller: formal public meetup or is it just, Hey, it's kind of a, everybody kind of get, has it been introduced, like referral based? Yeah. I

Shantanu Singh: don't know if other folks that are listening to this, um, you know, have lived in places like I used to live in Chicago, San Francisco.

Mm-hmm. If you hang out at a coffee shop long enough, you'll meet interesting and inspiring individuals. Yeah. Um, and so for me, those, absent those meetings that are not pre-planned, like are the best meetings and even if it's just to learn like on the ground level, who got funded, who, um, you know, who's doing what, um, what events are happening, you know, things like that.

Um, I used to live in San Francisco during the.com bubble in, um. 1999 to 2000 and, uh, I had just graduated college and I was living, um, in a neighborhood that's adjacent to the Haight Ashbury neighborhood called Coal Valley. And I mean, there were well-known folks that we know today that were coffee shop rats, like in that, in that neighborhood.

Um, and, and seeing that type of hustle and just remembering it, even to this day before, you know, you see folks on the nightly news or wherever they are, uh, it's inspiring because it all starts from something. It just doesn't magically happen

Jacob Miller: if you're, if obvi like, is successful and keeps growing, um, and you, and you decide to do it full time.

Like, will you stay in Wisconsin? Do you feel like you would move to a different hub? I'm kind of curious, what, what keeps you in Wisconsin?

Shantanu Singh: Why would I not stay here? Is one question. Like,

Jacob Miller: I'm just curious. Yeah, yeah.

Shantanu Singh: The why, why would I stay? Because I've had the good fortune to travel the world and live elsewhere in the United States, and I realized that when you're really focused on something, it doesn't matter where you live.

Mm-hmm. Like, um, maybe if I was 23 and I lived in San Francisco and you know, someone was like, Hey, you can do this in Milwaukee or San Francisco, where do you wanna work? Well, yeah, there's an obvious answer to that, but that does, that's not me today. Um, and I would love to grow this company in Milwaukee.

It's not just a matter of me staying here as like outta principle. It's more, it's more like, I think there's a lot to do here. Like, there's impact you can make in this community that will be very visible right away. Yeah. Um, and to me that's cool. Um, to me, like going to a larger metropolitan area where there are more dots to connect because you are laying more dots and other people are laying more dots.

So these networks are just going to naturally, you know, happen faster. That is true, but that doesn't mean the impact of that network is proportional to the size of it.

Jacob Miller: Yeah,

Shantanu Singh: yeah, yeah. Right. So like you can have a really tight network that's small, but the impact is larger.

Jacob Miller: Yeah. If, if anyone's curious about your, um.

AI feedback cohort. Uh, w is, would they be okay if they reached out to you, uh, and asked, just asked about it, or, yeah, I'm just kind of curious. Just wanna make sure. Obviously I wanna respect that private group and, and everything, but I also think it's good, you know, obviously you, you are all the filter for that group to make sure like, hey, this is a good person.

They have good intentions, they're here, they're not just here pitching and stuff. They actually are working on something interesting. Let's it like, have 'em be a part of it too. Um, how, how should they reach out to you if that, if that's, if they are curious? LinkedIn, just through

Shantanu Singh: LinkedIn. Okay. I go, I do LinkedIn.

I, I monitor that pretty closely. I'm active on LinkedIn. When I have time, I usually publish quite a bit and give them two week period. Then I'm crickets for crickets for a while and then come back. Um, sure. But, uh,

Jacob Miller: um,

Shantanu Singh: one of the, one of the things that I think folks, um. Can do in terms of self-help is if you have an idea, you should be able to prototype it in no less than three days.

Hmm. Like you were, like, you can turn Gemini on your browser and it will walk you through step by step what's on your browser. Like invest a 20 bucks if you can. And it's unlimited education and it's unlimited customer support. Um, yeah. And if you don't know how to use GitHub, it'll teach you how to use GitHub.

If you don't know how to do a commit, it'll teach you how to do a commit. If you don't know how to, no code, low code vibe, code, it will walk you through it. Mm-hmm. And write the code for you and then debug it for you visually. Like, there's all, there's, there's all sorts of ways to build stuff today. That's actually really, it's, I sometimes can't comprehend it.

Jacob Miller: It's, it's interesting 'cause I, I've been using Claude, uh, and GPT to help me, you know, do some custom code on websites that I manage, and it's like, okay, I wanna make this cool effect happen. Mm-hmm. Or I want this in interaction to happen and it's like, I wouldn't have been able to do it before mm-hmm.

With like, custom styling and all that stuff. And it's like, okay, now I can see it in real time as an example. Like, I prompt it up and okay, I'm looking to do this and this and this. I want to feel this way. And it shows me an example and I'm like, I can then give it feedback like, oh, I like how you did this, but I want this to be more like this.

Mm-hmm. And all of a sudden, and then I just copy the styling and bring it into Webflow and I'm off to the races basically. It's a, it's a, and until you experience it, it's hard to like how you're explaining this to people. Like, oh, if you don't know how to do it, it'll teach you. It's literally what it is, but like, when you experience it, it's, it's kind of unforgettable.

'cause you're like, whoa, I feel really empowered right now. Like, this is, like, last year this was not possible, like for me, you know, so to speak. Obviously there's been, gpt has been out for a little bit, but a lot of the coding tools and the way that the interfaces and your web browser has gotten so much better, I feel like, over the last year.

Um, and the speed of it, and I just, I encourage anyone listening to just give it a shot. Like just mess around with like, ask it to make something for you. Like, and it'll make it, it's just, it's pretty incredible. Um, uh, there's other tools out there too, like lovable and bolt and stuff like that. I don't have any experience of those yet.

I definitely am gonna be trying those out soon. Um, but yeah, definitely encourage people within three days. Is is a total true, true to true statement? So for a prototype?

Shantanu Singh: Yeah. And that's, and that's what I mean is like if, uh, yeah, I, you know. I just wanna help people, but I need to make the right bets and so does everybody else.

And one way to demonstrate that is to de-risk your ask. And one way to de-risk it is to actually show something, um, rather than, you know, say, Hey, I have this idea. Okay. When you have that idea, how do you think you could turn that into an actual piece of software, if that's your goal? Or if you have this idea, how do you think people would pay you to provide them services around that idea?

Or, I have this idea, what if you wanna open source it, do you know how to open source it? And just learning that through AI and the self-help shows like, okay, this is just not something that's a transient concept. This is something I'd like to build.

Jacob Miller: So you talked about your co-founder, you talked about these hackathons, and then you have this, this AI.

Kind of founder feedback group. Are there, is there anyone else that's been like, you know, uh, a collaborator or like a mentor to you that has kind of helped you on your journey? Like, especially maybe in Milwaukee? I'm kind of curious if there's anyone you wanna give a shout out to as a thank you that's kind of helped you on your journey, uh, becoming a founder at least.

Shantanu Singh: Well, I know they're private people, so, um, sure. I'll speak in general terms. Um, a lot of, yeah. I mean, I think my, my place where I am today is completely proportional to the number of people that have helped me out. Um, and I try to help them out in ways that I can, and you actually learn a lot more by trying to help someone solve a problem that's not yours.

Mm-hmm. Uh, in my opinion, because that gives you a greater tool set. 'cause invariably someone will say, well, that's not gonna work for me because of X, Y, and Z. Well now you just learned something and now you're also working with someone to help them. Um, yeah. So at different tiers. So I've had great mentors that are not in, uh, Wisconsin.

I also prosecutor in Chicago when I first, um, exited law school or, um, graduated. And so the judge who I worked for, who's now passed, um, he was a huge influence on me. Um, he was like a very inspiring individual and extremely funny. He was, he, he originally was from Wisconsin, um Oh, cool. But grew up on the south side of Chicago.

Um. As a, as a lawyer and as a resident there. Then I've had, uh, really great mentors when I was a prosecutor and prosecuting credit crisis cases. So, you know, the movie, the Big Short, all of that stuff where people trying to profit it off of, uh, my mentor and his team, which I was a part of, and other teams and other offices across the United States.

They, they were a great group of attorneys to work with. Um, I learned a lot about perseverance and a lot about, um, if you really believe in something and you've researched it and you know, everybody's persuasive arguments and the weaknesses of your own given argument. Um, that just gave me a lot of confidence really early in my career to be.

A lawyer that I hope people would get value from. Um, and then I grew up in the United States, hadn't been born in India, and so I have a close group knit of friends in the Indian community that I've known since I was five years old. And yeah.

Jacob Miller: That's awesome.

Shantanu Singh: So now one of, I, um, like the latter stages of 40.

Um, and so I'm still very, very close with them and I'm also friends with, um, other folks in the Indian community through my brother, through my family. And, you know, they've been fortunate to become CEOs of their own companies, exit startups that they founded. So. It's a lot like osmosis. I can just call someone, and if they're not, they'll definitely gimme their time and tell me, you know, not so much telling me what to do, but like considerations that someone should think about at this stage.

Jacob Miller: Sure. I'm not gonna answer your question, but here's questions to think about. Yeah, yeah, yeah, yeah, that's great. Um, as we wrap things up, what's next for Obviate? Um, anything coming soon that you're excited about or you just say, Hey, we're heads down working on the product, getting customers. What's next?

Shantanu Singh: Uh, what's next for us is we launched a, uh, prompt engineering, um, curriculum.

Um, and so we've had success, uh, speaking to in-house lawyers and teaching them about prompt engineering. Um, so that's something that we're excited to grow. Um, second is, um, we're getting some really great feedback on some of our new features. And that's what we're always looking for, is folks to use our tool so we can learn what needs to get better.

Um, in fact, right before I got this call, we got some feedback and it was pretty, pretty great feedback, so I was really happy to get it. And third is we're just coding heads down and we're in a place by end of the summer where we're just going to start really commercializing our offering based upon the learnings that we have over the next six weeks and over the past, you know, eight months.

Jacob Miller: Yeah. Awesome. Well, congratulations on all your progress so far. Best of luck on what's next, and, uh, thanks again for taking the time to talk with me today and, and sharing your story. Really appreciate it.

Shantanu Singh: No problem, no problem.

Jacob Miller: Thanks for joining us on the Startup Wisconsin Podcast. Wanna support the show.

Don't forget to subscribe and get updates. If you're feeling generous, you can share, rate and review our podcast to help others find us. Alright folks, until next time, let's keep moving Wisconsin forward.

I.

‍