How a Silicon Valley Lawyer Accidentally Created a $3B Unicorn

So the very first version of Ironclad was an email alias, admin@ironclad.ai. And if you were doing a contract, you would CC that email address. It was just me doing all of the stuff and we were like, well, eventually we'll make this AI. For now it's just me. There were some funny moments in there where I had to get off Caltrain to respond in a timely manner as Ironclad. That was obviously not scalable. So I think we realized that one pretty quickly and AI was nowhere near being able to make that idea work.

The Library of Alexandria was an attempt to capture all human knowledge.

Fire and war destroyed it.

Today, we try again. Welcome to the Library of Minds, where you'll learn the frameworks and mental models of the world's greatest thinkers. And when our conversation ends, yours begins. Each guest has created a digital mind on Delphi that you can talk to anytime, anywhere. I'm Dara Lajevardian. Let's dive in.

Thank you.

Thank you for giving the idea and for coming. I think the way that we met was we were sitting next to each other at Ascent and Jess texted me, holy shit, the guy next to you is so impressive. You need to talk to him. He was in AI in 2014.

Yeah.

But I think before we jump into some of the specific questions, is there anything like Talk us through your path in Ironclad just to set the stage for our conversation.

Yeah. Well, great to be here. I'm excited that you're doing the speaker series and it's really an honor to be here with you guys. It's exciting to be in just this environment. I mean, you're really lucky to be here right now. And when I think about my journey at Ironclad, I really look back on this almost exact time as some of my favorite memories of the company. I know there's lots of really hard stuff that you're going through and there's lots of challenges with this time, but they're fun challenges at the end of the day, and I think you all look back on them really fondly. So my journey with Ironclad, I was a law firm associate. I didn't plan on starting a company and I started basically doing automation hacks for myself at the law firm I was working at, followed that train down, eventually kind of became this little bit of an expert at my firm on building productivity tools internally. I ran the evaluation of IBM Watson, which if you guys remember that Watson won Jeopardy, I think in 2013 maybe. And so there was a lot of enthusiasm about AI, which is how we got to the ironclad.ai. Founding back in 2014, quit my job, met my co-founder, and I'm sure we'll have a chance to talk about other parts of that journey, but that's kind of a little bit of how I got started.

Yeah, yeah, yeah. I think—

Help set the stage.

Definitely. And I think there's also two interesting things that have happened recently. One is that you just stepped back from being CEO after 10 years.

Yeah.

And maybe talk about also just an explanation of what Ironclad is and where you guys are right now in the journey of the company.

Yeah, so Ironclad is a contract lifecycle management platform, CLM. And what that means is we help businesses of scale with the whole lifecycle of their business contract. So everything from the idea that a salesperson wants to enter into a relationship with a customer or that the company wants to hire an employee or that the company wants to buy a piece of software. All of those are ideas that result in contracts. And we help those people turn that into the right template, the right approval chain, the right signature process and logistics. And then we track all of the data in those contracts at the end of the day so that when someone needs to act on the information in a contract, they've got it in the system that they need to take that action in, or they can easily pull it up so that they can find it. And that's important for a company like you guys, but imagine some of our customers like L'Oréal or Mastercard or Heineken that are running global supply chains and have millions of contracts. It's essential for them. So that's kind of what Ironclad does. And we have AI sprinkled on top of that from everything that helps with contract redlines and the diffs between versions of contracts to extracting data from PDFs that might be 30 years old in some cases.

That's a problem for our content extraction system is getting well-structured data out of the data that people give us. I think one of the things that you said to me that made me realize, holy shit, this is how the scale of this problem is. You were like, think about Walt Disney, think about the hotels and the people they hire for the amusement parks and everything, the TV shows. Imagine how many contracts there are. Like, how do you even know what you've agreed to?

Yeah.

So I think that leads well into the thing that we were just talking about right before we started, which was an interesting strategic decision that you guys made early on. And I think you even in an Instagram reel called this category expansion play recently. Can we talk through maybe some of the, like how you arrived at this?

So the full like unpacking of how we got to that. So CLM is a big category and like I mentioned, started with AI and just this kind of idea that Hey, if Watson can win Jeopardy, Watson can probably help, or technology like it, lawyers with all of the complexity of their work, long-form text, dealing with complex information and finding a fact in a large body of documents. So the very first version of Ironclad, which was the truly hacked-together version, was an email alias, admin@ironclad.ai. And if you were doing a contract, you would CC that email address. So you'd be like, hey Sam, I want to do an NDA with you. I've CC'd Ironclad on this email transaction and the system will help us get the NDA in place. And Ironclad, which was actually me, Ironclad AI would be like, hello Sam, we need the following information from you to do the NDA. And Then you would get this DocuSign request and I would fill in the template and it was just me doing all of the stuff. And we were like, well, eventually we'll make this AI. For now it's just me. And there were some funny moments in there where I had to get off the train, Caltrain, to respond in a timely manner as Ironclad to do contracts, get back on the train. That was obviously not scalable. So I think we realized that one pretty quickly and AI was nowhere near being able to make that idea work. But we realized, well, the thing doesn't need to be on an email thread. Actually, people would probably prefer to use an interface here, especially at scale where they can click in and they can fill out some information and then Ironclad will fill in the template, right? And then when Ironclad is ready to get signatures sent out, we know that right order to send them in. And since we filled out the template and we collected all the signatures, we actually know a bunch of metadata about the contract anyways without even doing any data extraction. So we can start tagging some things. So basically the initial version of Ironclaw was really good contract workflows and contract workflows wasn't really a category. I mean, we could sell it. I would go to Dropbox and I would say, Hey, what contracts do you guys not like doing over and over again that are really repetitive? And they would say, well, our NDA. Yeah, we'll pay you $500 a month to get an NDA workflow. We'll pay you $1,000 a month to do a sales contract workflow that our sales contract people could do. So we were kind of just like slinging contract workflows to anyone that would buy them. It wasn't really a category. We would often have to explain what we do. There was enough of a problem there that people would buy it though and pay us at least some amount of money. And then I remember getting an email. We had hello@ironclad.ai on our website, and I remember getting a literal one-line email one day that was, hello, are you a CLM? And I had not heard of CLM before, and I almost deleted this email. It's actually crazy to think about What would've happened if I deleted this email because it was so influential in our company trajectory. And I Google what is CLM, contract lifecycle management, and it basically describes a database for keeping track of your business's contracts. And I'm like, well, we don't have a database. I mean, we have all these contract workflows that we've sold. We do have some information that we could pretty easily use to populate a database. And then after I Googled it, figured out what it was, I was like, let me see who sent this email. And it, it was apigee.com, which was like a pretty small cap public company. Google has since bought them. And I was like, okay, well this is from a public company that wants something called CLM and it like wants it bad enough where they're emailing us who has like a really crappy website, one page up that's very vague. And asking us if we're a CLM. So of course I responded, yes, we are definitely a CLM. Would you like a demo? And we booked a demo for that Friday or something. And then I told my co-founder, I was like, hey, we're a CLM. You're going to need to code up a database. He's like, dude, what are you doing? Why are we doing this? I have to take the train down to San Jose. And I was like, oh, it's okay. You can just code the database on the train. We'll do the demo. We'll see if they like it and I'm sure we'll be fine here. So we take the train down, he codes up the demo on the train, we do the demo and the person is like, wow, I've evaluated 25 different CLMs. I've tried to find everything I possibly can. I love this because you guys have contract workflows and everyone else has really bad workflows or they don't have them. So we eventually go through this process as a really small company, get this account as a publicly traded company, and all of a sudden we're a CLM. So that's the in the weeds answer, but maybe some relevancies in there to how we got from AI to workflows to now our category.

Yeah.

And that took about 2 years.

I mean, the thing that happened, so when you were like, when we sat down, we started talking about this, one of the first things you said to me was, you guys should think about positioning yourself as a CRM.

Yeah.

And I just had this instinctive reaction like CRM, we're not a CRM. And then he told me this story about how it was kind of the same where you're like, we're not a CLM. Like we're so much more than a CLM.

Yeah.

But, for some of these people and a lot of customers, you have to bring them along the journey. And so you have to put it in terms that they understand.

Yeah. So like CLM really was just database software then, which is why we had the same reaction that you had to me saying CRM is we're like, oh my gosh, this is like super clunky database software. No, we're not that, that is really lame and we are so much better and we'll eventually disrupt that. But what we realized from that publicly traded company suggesting the term to us was it did open the door. One, the customer wants to buy a CLM. You don't need to explain to them that they don't want to buy a CLM. They want to buy a CLM. And just saying yes to that allowed us to opt into so many, before we had to create the demand almost for contract workflows and people didn't know that they really had a contract workflows problem. When they sh— when you showed them solution, they were like, oh, I can use that. But now it was like there was hundreds and thousands of people out there looking for the CLM to buy and we could go in and say, hey, we're CLM, check out this. Like everybody, and we actually use this position where like everyone has a database, not everyone has contract workflows. And our database for many years was like truly bare bones. And I would say worse than other CLMs, which is the main feature of the category before we started competing in it. And we have now shifted what a CLM is to be contract workflows. Like in 2025, you could not win a CLM deal unless you have contract workflows, um, which we, we remain the best at. But it really did shift the market and kind of expand the market to a whole new set of features that weren't considered part of the category before.

That's such an interesting progression of 10 years where at first it's just a side feature and then you just become so good at it that in 10 years people expect that.

Yeah. And I really think there's something to, in some ways it's like bundling. So when you bundle workflows with a repository, like where those things live after the workflow, the sum total is greater than the individual parts. And I do think with AI, there's a lot of power to that general bundling strategy.

Yeah. Well, so I think this leads well into the next topic, which is Another parallel between Ironclad and Delphi is the fact that trust is such a foundational component of do I trust you to do all of my contract management? And same for Delphi, put all of my content into Delphi. I want to feel as if it's a trustworthy platform. And so how have you, what has been your thought process or prioritization framework heuristic for trying to balance Time to market, speed, do things quickly, but we cannot violate customer trust.

Yeah, I think it's really finding a balance on a lot of different factors. I mean, obviously there's the startup truism that you move fast and break things, which is up on posters and walls, and you can't move fast and break things in the legal industry in particular. You have to kind of move fast and not break things. So we invested really early on in quality engineering is one technical decision that we made. So actually I think our second hire, maybe our first hire outside the co-founders was a quality engineer and she was a really early, actually she was the first quality engineer hired at Palantir. And she basically, her superpower is breaking things. So she once famously got on an airplane and started putting in stuff on the in-flight entertainment system and broke the entire flight's in-flight entertainment system. She just has the mind of, she's like, I'm going to break whatever I'm looking at. And having someone like that internal to the company. Where does she work right now?

She's still at Argo. Okay.

Yeah.

So, and that's what she does. She's like, how can I break this before our customer breaks it? And she'll report that back. So she's actually the old school version of quality engineering and we have other teams that are more test writers. She is break the software.

Interesting.

That's her full-time job. So we did some stuff like that. I actually think this is maybe counter to some advice in Silicon Valley too, but branding and naming was really powerful for us. SpringCM was a much larger company than Ironclad at the time. And would you rather have your contracts in SpringCM or Ironclad?

Ironclad's cooler.

Right? Yeah. It feels safer, it feels more trustworthy. And then we oriented our brand colors and everything around trust and we looked at the way trusted entities like banks and law firms brand. And then we flipped that into a more modern context, but we used a lot of those colors, but different tones of those colors. We built in some cultural stuff too early on. I'd say there's a theme at Ironclad of we want to appear to our customers as if we're a bigger company than we actually are. A funny anecdote around that was this Apigee company that was our first kind of enterprise customer. I think we signed them up when we were maybe a 3-person company, and then the company had grown a little bit as we continued to work with them. And the first, they closed their quarter as a publicly traded company doing all of their contracts on Ironclad. So like all of their sales contracts. And I remember they were so happy. They were like, hey, we want to do a happy hour with you guys. Like our global head of sales is flying in. In from New York and the whole sales team's gonna be there. Like we're gonna do an Ironclad and Apigee happy hour. We get there and I think the company's maybe 5 people at that point. And so our entire company is there and it like talking to their salespeople and you could just see it dawn on people at Apigee that like, oh my God, this is a 5-person company that we're working with. Like they clearly thought that we had like hundreds of employees and stuff. And then they like, wait a second, you do all of the software engineering, you do all the quality engineering, you're the CEO. But we obviously, they trusted our solution. There was enough in the way the product showed up, the way they had interacted with our team, the way we had handled ourselves and interacted with them that we were conveying trust. And I do think that's important.

Want to go deeper? Talk to Jason's digital mind on delphi.ai, link in the description.

Another thing that you emphasize heavily in some of our conversations is just the talent density of the early Ironclad team as being so critical to success. And I'm wondering, you've had to interview so many people in 10 years, what has been or what is an interview question or questions that unintuitively give you a lot of signal?

My favorite one is I try to subtly goad someone into complaining about a previous job that they've had. And basically if I can get them to complain, then it's instant failed the interview because I just feel like if you're going to, and the people that are gleefully, some people will just gleefully go into complaining about their former job. I don't mind a little bit like, yeah, that wasn't great. That's fine. But if I catch that they like complaining about their previous job, that's the insta-fail. Because it's a startup, there are just going to be so many things you could complain about. And I kind of feel like this stage is you have to find the one thing that's working in the 99 things that aren't working. And so if you like complaining about stuff, You're gonna focus on the 99. And what you actually want is people that are like, get really locked in when they start to see that, that one thing, and they're gonna make that one thing great and they're gonna get excited about that and they're gonna kind of be able to put on blinders on the 99 things that suck. So that, that's probably my one thread that I've carried through the years. The one that I picked up later, which I, I think is maybe it's more important later, but I've realized that I personally don't work that well with, um, the negative way of saying this would be like people that are insecure about their ability. Um, and I, and like need positive reinforcement about how they're doing. I think that's more of a personal chemistry thing of like, look, if you're doing a great job, I'm going to tell you you're doing a great job. But like, if you're just doing your job, let's focus on winning. I'm not going to like be negative, but it's just not who I, how I really show up that much. So I've kind of learned that that isn't as good a fit for me personally.

How did you go about evaluating Dan, the new CEO?

Well, Dan, yeah, so our new CEO, um, I, I, someone I've known for a while because DocuSign's kind of an adjacent company to Ironclad. Like we have a great integration and we're partners. Um, and it, I'd just gotten him to, to a few touchpoints with him over the years and I'd always thought he was a really thoughtful person. I love that you guys have this library here. People at Ironclad love to read and talk about weird stuff. So Dan, I remember the first interview that I, in this role, he threw off this Tolstoy quote from Anna Karenina in his first interview offhand and I was like, that's a really good fit with Ironclad. That's just our culture. And so keeping an eye out for little things like that.

Yeah.

And so one of the unique things that as I was doing research for this, I found in a bunch of different things is the fact that you read poetry to the team.

Yeah.

And so at the start of every year you read a poem.

Yeah.

And I'm wondering, is there any particular year or poem? Because I imagine they're motivated by whatever's going on or a lesson that you want to convey. Is there a particularly memorable one? Yeah.

So my favorite poem is that I read to the company and is one of my favorite poems is Ozymandias by Shelley. And this has particularly special meaning to me. So I got really lucky. I know there's some Notre Dame people here. I did my undergrad at Wake Forest, which is in Winston-Salem, North Carolina. And Winston-Salem also happens to be where the late poet Maya Angelou retired. And in part of her retirement, she taught one course at Wake Forest every year or something. And I think there was 15 people in the class. And so you had to lottery into this. And I just remember being like, I was not interested in poetry at all. I thought it was dumb, but I was like, ah, I could potentially take a course from Maya Angelou. That's cool. I throw my name in the lottery. And for some reason, I get a slot in this class, and it was like the definition of a life-changing class. And I will always remember the first day that she walks in and just before she says anything, she just recites in Latin this quote that translates, "I'm a human being, therefore nothing human can be foreign to me." And it was written by this Roman poet Terence, who I think like went from being a slave to kind of like a literary figure in Rome. And the whole course was about unpacking this concept of no human experience can be foreign to you. And her way of having you experience things was through reading poems. So she would take someone who had grown up in North Carolina, had never left the state, And she would give them a Robert Burns poem about a red rose in Scotland. And she would have them really try to inhabit the character of Robert Burns, walking around a lake in Scotland, finding a red rose, read it in a Scottish accent, all of that stuff. And she would select a poem for everybody. The poem she selected for me was this poem, Ozymandias, The visual description of the poem is there's a traveler in an antique land, walking through the desert, and he comes across basically two legs of stone in the desert and a shattered face next to it. And he brushes away some sand from the pedestal, and the pedestal reads, "My name is Ozymandias, King of Kings. Look upon my works, ye mighty, and despair." And then the last line of the poem is, and the lone and level sands stretch far away. So basically there's this shattered statue in the middle of the desert with this look upon my works, ye mighty, and despair. And I read that poem to the company and I talked about what does it mean to be a great company in the face of that truth that our company will die, we will die. So what does a great company mean knowing that it ends in just like that statue did. And I think the definition of a great company, which I talked to Ironclad about, is a company that changes the trajectory of an idea even after the company is gone. So the two examples I gave at the time, which were Tesla for electric vehicles, like my garage of my apartment building over COVID went from from like one electric car to basically being probably 95% electric cars. And Tesla changed the sort of definition and trajectory of electric vehicles. So I think Tesla is a successful company. I think that's an interesting way of thinking about like what a successful company is, that even if that company ceased to exist, the idea itself is different.

Yeah.

So a little bit of a long-winded answer, but you asked me about a topic I'm passionate about.

No, that was fantastic. I again see parallels where something that we're hopeful for is that DALL·E is going to change your trajectory of what it means to capture and preserve and transfer knowledge.

Yeah, and I believe that. And I think the fact that you know that and can answer that is very unique. I appreciate it.

And so I think we just touched on a bunch of interesting formative experiences you had prior to the company, during the company, Has there been a value or a lesson from upbringing or early career that you have realized has been critical and maybe one that you had to unlearn?

I think upon reflection, I think taste is really important and that's something we inadvertently got right in our early hiring and early team. Taste, if you have taste, you're going to make the lots of little decisions that you are going to have to make on a daily and hourly and even down to the minute basis. So many little decisions that people at the company have to make and y'all are so consequential to the future trajectory of the company that little things you do now that are not part of your job, but you just take upon yourself to set up the support ticketing system. Someone's probably made a decision on that already and someone may regret. Yeah. But you have to, and the the messaging that you put into the support ticketing system and what's the first message, what's the email? There's just so many little things and you're gonna, hiring people that have some shared sense of taste goes a really long way. And I can actually see that you guys do have taste. The office is cool, but it's not too cool. It's the appropriate, there's lots of tasteful things that have happened in getting this together, the idea library, that's so cool. I love that. And I think holding onto that and being intentional around the taste that you have and looking for that in future teammates is important. And that should be everyone at the company in my opinion.

I love that answer. I feel you were early to that as well because now everyone's talking about how taste is the only thing that's going to be around.

Oh really? Yeah. Cool. I actually didn't know that.

Yeah, this is a thing that people, the idea people on Twitter have been talking about recently that In a world where execution is abundant with AI, the only thing that matters is taste.

Cool.

So there you go. AI in 2014 and taste in 2015. Cool. So I think maybe one or two more questions and then we'll open it up to everyone here. I think one of the things that I'd also be curious here about is, and I think this goes in relation to taste and also speed and trust at different threads we've been pulling on is sometimes you have to make difficult decisions about saying no to things. Yeah. And I'm wondering what has been maybe a memorable time where you had to say a very difficult no and what was the thought process?

We had to churn a lot of our early customers, which was really hard and it felt really risky at the time. So basically the first contract workflows we ever made were for YC companies. So we coded together this system that it's still in use today, hilariously, that is the system for YC companies to do SAFEs. And we coded that up a couple hours before demo day. And So that was kind of our first success as a business. We just powered Demo Day after coding that, and then we're like, okay, well we should do this for all YC companies. So we went on and signed up, I don't know, 300 YC companies doing initial contract workflows. And that was exciting because month over month growth was great. Usage was up. Our friends had organically used our product. But eventually we realized like, one, actually a lot of YC companies go out of business. So like your churn is relatively high if you're signing up like the 2-person version of a company that might pivot their idea like 6 times. Like a lot of those go away. They're also super price sensitive. Like we would have people like upset that we raised our price from like $100 a month to like $125 and they're like, I'm churning, this is ridiculous. 'cause they're running on ramen budgets and stuff. So we just realized this isn't a great business to be in, selling $100 a month to YC companies, even though we are getting some growth on it, it feels kind of good. And so this was like, we had this big tension between the couple businesses that we'd signed up that were enterprises like the Apigee but we didn't have enough of them to like be confident that that would work yet. And then we had these hundreds of YC companies that were kind of like taking up most of our daily time. And so we had to kind of like shut that down and it was painful and scary and it resulted in like, I think the only time in company history, which is a few months time where our like revenue went down. So all of a sudden we went from this to like this. And it was scary and I think it was scary for the team. It was scary for me, but we sort of bridged that gap eventually and the new momentum came in when we cleared, we said no and we cleared the space to work on the thing that ended up working.

Okay, so I think maybe we'll do rapid-fire questions. I have 4 of them and then we'll open it up. So I think First, have you made a decision that seemed irrational to others, but in hindsight was actually right?

So working on legal was considered a small market. Ironclad, 600 employees, I think there's 5 people on our legal team. So people, investors were like, why are you going to do this? That's just not a large team. We have hundreds of people in our go-to-market organization, our engineering organization, and investors are like, why would you sell to the small team instead of the big team? And then furthermore, we decided we were going to focus on legal operations, which is one person. So at Ironclad, 600-person company, one person, but being really focused ended up helping us. That was longer than a rapid fire.

Thing you will miss the most about being CEO?

You have a lot of agency. I mean, you you can change anything. Even with the board. Yeah. I mean, it's more so than, and I will, that's the thing I'm adjusting to the most where I'm like, oh, I can't say yes to that anymore. I need to try to follow the process internally and be a good citizen. So I'm adjusting to that.

Favorite memory from Ironclad Series A.

I mean, even if you remove the Series A, the That's the thing I was saying at the beginning. I would go back to this time. We had an office that this really reminds me of and we would just do fun stuff like this and we would have law professors in and they would talk about some legal concept to people at the company. And I think if you went around Ironclad of people that were there at the time and asked, they would just pick a random day around this time of company.

Cool. And then the last one is why are you excited about Delphi?

Yeah, I mean, I think you are going after a really big idea. And I like that. Like you're going after something really big and if you're successful, which I really believe you will be, the world is different and the concept of not like how we think about knowledge as human beings is different and changed. And I love the historical reference in your name to that.

Awesome.

Any questions? Yeah. First of all, thank you for coming. I agree with you. Today is a very exciting period to be in AI, but you were able to build a business in a time before people turned irrational, before AI had so much hype.

Yeah.

Are there any lessons or principles they would like to share with Delphi so we were able to like really find signal in all the noise that is going on today?

Yeah. I mean, this one sound, this one is gonna be uncontroversial, but I just think it's worth hammering home because there are decisions that all of you can make that will concrease the surface area of the company here, which is spending time with your customer and going back to the Maya Angelou thing of deeply understanding your customer. I would want to know for our early customers, what do they read? What does their day look like? Do they have a long commute to the office or a short commute to their office? Do they like working on their phone? Do they not like working on their phone? When they're not at work, what are their hobbies? I mean, we would just want to know everything about them and we would turn nuggets from that into our pitches even. I remember a customer saying, the most important thing for me is to finish my work so I can go home and spend an hour with my kid before they go to bed. That became a big part of the Ironclad pitch. You can actually take some, and I love working on my phone. Whenever I can approve something on my phone, I love to do it. But the way that PDFs display on my email client, I can't get them. All of a sudden we're going from, hey, save 5 minutes on a contract workflow to you can spend more time with your kid, like actually true. We got that from deeply understanding the customer and less of like, do you like this interface or this interface? And more like, what's your day look like? And I think everyone at the company can and should do that and pass that along. And the coolest thing about this stage, you can all talk to each other. You have a customer conversation, great. You're going over the person that built that feature, found this about. You're going up to someone who's selling, found this. And we are not going to do anything that's not scalable, but we are going to make sure that we deeply understand what they're doing and we're going to not have any rational reason for our competitor to go with that company. And there's been 2 or 3 moments where we kind of put to bed some competition there. Later stage from where you are though.

Really grateful for you coming and talking to all of us today. Thank you.

It's been really fun. Thank you.

Thanks for tuning in. For personalized advice from Jason, head to delphi.ai and ask his digital mind.