0 → $1b in a year: The remarkable story of Handshake's refounding

**Garrett Lord** (0:00)
I remember calling Mamoon and Margo and Will and Megan on our board, literally over the holidays. I laid out a whole investment thesis. We wanted to build a target of $25 or $50 million of run rate revenue. That seemed insane. I remember asking Will Reed, what do you think good looks like in a year? And he's like, I think it's probably $100 million. And I'm like, dude, you're absolutely crazy.
I mean, we're starting from zero, we have one lab relationship, and now we've gone zero to a billion in a year. We're probably one of the fastest growing AI companies out there. The opportunity in front of us is so big that it just requires incredible execution. How can we be more ambitious? How can we turn around on a month plan in a week? That's basically my entire job. I spend 0% of my time celebrating any of our success.

**Nakul Mandan** (0:41)
Garrett Lord spent 10 years building Handshake into the career network for an entire generation. 18 million students and alumni, 1600 schools, almost every company in the Fortune 500 is a customer, and it's a company valued well north of $3 billion.
Most founders would spend the next 10 years protecting that. Instead, over the course of 2025, Garrett did something almost no founder does. He refounded the company around a business that was barely a year old. Handshake AI went from to a billion dollars in gross revenue in 15 months. And to get there, he did layoffs, reset the entire company, and bet the thing he'd spent a decade building. Today, we get into the why and how of this transformation, and how Handshake is shaping the future of AI. Nakul up.
Garrett, welcome to the show.

**Garrett Lord** (1:36)
Yeah, thanks for having me.

**Nakul Mandan** (1:37)
So Handshake started in 2014 as a LinkedIn for college kids, and over the decade ahead, you built it into a $3 billion business. Yet in 2025, you decided to transform it completely into a new thing, Handshake AI.
Let's start with the genesis of Handshake AI. What were the early days like and what gave you this hunch that you need to go in this direction?

**Garrett Lord** (1:59)
It all really started at a Christmas party, where I was talking to a researcher at one of the frontier labs. So much of being an entrepreneur is I think trying to talk to customers, understand trends in the marketplace. I had seen quite a few recruiting companies really focused on AI and really serving the labs. I was particularly curious. I mean, I think that's one of the main attributes, a lot of entrepreneurs is just trying to figure out what's going around you. I use that opportunity at the Christmas party to dig in with this research. It was actually kind of funny. Normally, at a Christmas party, you meet 25, 30 people and I basically talk to one person all evening, just trying to understand a little bit more about what they were recruiting people for.
What was clearly articulated from him was that there's really three stages of model building. There's pre-training, which is scraping the Internet. There's mid-training, which is called supervised fine-tuning or feeding in high-quality samples. And then there's post-training and post-training is really around creating environments where models can learn from reinforcement learning, really codifying a lot of the knowledge and judgment that wasn't on the Internet.
It became clear in my conversation with him that obviously there can be continued gains in pre-training, but a vast majority of the gains that were going to be had in modeling was going to come from digesting what was in people's heads into forms of fuel in reinforcement learning that models could learn from. He was articulating this massive build out of post-training and the amount of data that is needed, and really summarize the conversation like there are three stools to AI. There's algorithms, there's compute, and there's data. I actually started our relationship two weeks later with that same exact researcher at that top tier lab, where we started working with them on recruiting STEM PhDs.

**Nakul Mandan** (3:41)
Just put a timeline to this. This was December 24?

**Garrett Lord** (3:44)
December of 2024
Yeah, exactly. Yeah. So much of being an entrepreneur is like building a small strike team, taking some of the best people in the company and starting to explore this idea. We started to reach out to a bunch of other relationships they had with other labs, and our real focus was on serving the top two model builders. Like that was, we thought that if we could serve these top two model builders with these frontier capabilities and kind of evolve with them, that that was really an indication of what the future would look like. And so it started off kind of as a small exploratory strike team, and very quickly by January of that year, this was my full-time job at the company.