Holding onto Group Project Energy (GPE)
This is a personal piece about building a startup, pieced together from moments with early investors and team members. A year later, I’m revisiting it for Substack, trying to capture what it felt like at the time. Names and moments are real; the quotes, as memory allows. I hope you enjoy the read as much as I enjoyed the write — Edward
"How do you know it’s a good idea to invest in a 22-year-old in a college dorm room?" someone asked.
In front of everyone, Jones responded, "You don’t know."
And that was exactly the kind of blunt honesty we needed.
We’d invited him to Montreal for a week-long retreat with the GPTZero team because great teams need honest conversations—about what’s working and what’s not.
At that point, we were seven team members—soon to be eight—split between our Canadian and American offices. It was the moment things started to feel real.
After just two Zoom calls and a phone conversation, The Tripp Let’s F Crush It* Jones** made a $3.5 million bet on GPTZero in March 2023. But that was just the start. For the first year, we checked in every other week.
Ever so often, he would ask me, "What are you going to be when you grow up?"
Grow up? At first, I thought he meant me. Then it clicked—he wasn’t talking about my age. He had invested in what was, at the time, a school project. His gamble was that it would grow into a real company.
And real companies? They have real-company things: a mission statement, bi-weekly sprint planning, monthly roadmaps and retros, and quarterly objectives and key results.
The Hackathon Spirit
In Montreal, we started working with team members on hackathon projects. For most of us, it was reigniting a feeling – a contagious, fizzing energy of making something new together.
Group Project Energy [G.P.E] had fueled us in the early days of the company. But as we grew, it had started to fade.
Many of us—Alex, Olivia, Jacob, myself—got our first taste of product-building through hackathons. And through a strangely specific niche: Chrome extensions detecting fake news. At one point, we joked that our hiring bar should be: "Has previously built a Chrome extension that detects fake news at a hackathon." A surprising number of GPTZero team members have.
Our team loves hackathons because they bring a necessary jolt – a reminder of what makes us different.
This isn’t Big Tech. This isn’t corporate. This isn’t a place where ideas get stuck in meetings for months before they’re built—or worse, before they’re buried.
A hackathon doesn’t allow for that luxury. It is a space of immediacy, where decisions get made in a half-second, and creativity flourishes not because there’s time but because there isn’t.
For example, award winning hackathons include ideas like a guess is this AI game, a GPTZero ‘human writer’ badge for Substack writers, and a GPTZero twitter/X bot to fact-check any tweets
And then there are the moments that crystallize why we do this.
During a summer hackathon in Montreal, as the clock edged toward midnight, Jacob Tian—our summer research intern from MILA (Montreal AI Institute)—suddenly stood up.
"Wait," he said, and without further explanation, disappeared out the door.
We assumed he had gone home for the night. Instead, an hour later, he returned—with industrial-grade lighting equipment. The kind used on professional photo sets. One by one, he arranged the lights around the room, switching them on.
The effect was immediate. People who had been drifting toward sleep were now upright, blinking under the artificial midday. The hackathon, clinging to its last reserves of energy, jolted back to life.
This is why we run hackathons. Not just to build, but to recreate that flickering, electric moment of Group Project Energy. It’s the best feeling in company-building—and maybe the rarest.
The participation awards were just as memorable—like the $100 Tim Hortons gift card (for our Canadian teammates) and the 3D-printed GPTZero company trophy. Because in some sense, participating in hackathons is part of the journey—taking ownership of a product, building something from nothing.
The Company Spirit
OKRs—Objectives and Key Results. That’s what real companies do. The framework, originally developed by Andy Grove at Intel in the 1970s, was popularized at Google by John Doerr in 1999 when he made it a core part of their management and execution strategy.
It now grounds our execution. We’ve already seen how it can keep teams aligned and make sure everyone is working toward measurable, ambitious goals. When done right, it can turn strategy from abstract ideals and conversations into tangible progress.
But great companies aren’t just built on frameworks. The best parts happen in between: on the car ride back from team karoake or spontaneously on a sticky note.
In 2020, I was working with Professor Thomas Griffiths, and Bill Thompson—now at UC Berkeley—on modeling modern conflicts over time using Hidden Markov Models. One night, while commuting from my co-founder’s apartment, an idea struck me: we could use the same models to predict which sentences in a piece of text should be highlighted as AI? That flash of inspiration became the first-ever AI detection ‘sentence-highlighting’ model. We deployed it, and the very next day, demoed it live on CNN—Anderson Cooper.
A year later, while reworking our models, my co-founder turned to me and said, “Edward, that was actually... pretty clever.”
At a team hackathon in Toronto, I patched together a PyTorch script to analyze our database for the most frequently used words by AI writers. With help from our ML engineers and designers, this became our new AI Vocabulary feature as seen on Forbes. In another hackathon, ML engineers Edwin and Nazar developed human-readable explanations for why sentences appeared AI-generated. In Montreal, our ML lead Alex Adam and intern Jacob Tian created a winning project: 2D visualizations of our text database. It was eye-opening to see how ChatGPT-generated texts clustered near other ChatGPT texts—and the same for human-written articles.
While OKRs keep the train on track, and execution aligned, without G.P.E. there is no fuel. Many of the best ideas at GPTZero were sparked on the car ride back from team dinner, or under the industrial grade lights in Montreal. It’s why we run hackathons as a team – to tap into group project energy, and preserve it.
Capturing the Moment
In Montreal, Tripp reminded us to treasure this journey. We paused to take a mental picture—grateful for the small moments that come with building something real.
Back from the retreat, standing in the elevator of our WeWork in New York City, I thought about it again.
Group. Project. Energy. Without that, even the best intentions can feel hollow.
"Bouncy yoga ball to sit on in office?" someone asked.
"Yeah, of course," I said.
Great companies might be mostly built through structured execution—OKRs, roadmaps, organized planning. But their ethos? That’s built in the moments in between.
GPTZero in Montreal, circa 2024
It’s a personal goal of mine to do more writing this year :) Thanks for reading this post, and don’t hesitate to reach out or connect on twitter or linkedin — Edward
Loved this - similar origin story from a different angle
Love to connect and collaborate on a couple of chicken eggs I’ve hatched