The Anti-Slop Approach to AI and Startups
There's a type of entrepreneur who believes they can build a business with a single click, as if it's an effortless endeavor. They buy a course and expect to have the secret recipe for a million-dollar side hustle. We're seeing an anti-pattern with AI right now, where people boast, "I have 50 AI agents running my company," or "I have a suite of AI that tells me what to do every day."
This is the wrong approach. You're just creating a slot machine—or more accurately, a "slop" machine. Everyone on Twitter claims, "I automated everything. I work one hour a week and make $10 million on the weekend with my SaaS app." Most of that is just marketing; they're lying to you. It's easy to sell hope and a vision of skipping the hard work, but shortcutting directly to the value will never happen. It'll never work. The only person making money is the one selling you that course. They've found a way to print money.
My name is Max, and I'm the founder and CEO of Gum Loop. Gum Loop is an automation platform that handles about 4 million workflows daily for huge businesses like Instacart, Shopify, DoorDash, and Gusto. We're a team of 15, scaling quickly. Our platform lets the people who truly understand a business problem—the marketer, the salesperson, the ops person—automate their own work without having to spec it out for an engineering team.
Rule 1: Throw Yourself Into the Ether
I studied software engineering at McGill in Montreal. I put a lot of effort into school, trying to get the best grades possible. My goal throughout college was simple: get a good job. I wanted to work in Big Tech, thinking that was the right path. I realized pretty quickly that I hated it.
Many people say, "I'll just go to Big Tech for a little bit, learn a bunch, and then do my own thing." I think that's a form of cope. A lot of people say that, but then they get trapped by golden handcuffs and never leave to start their own thing. I don't think I've used anything I learned in Big Tech in my startup at all. The only thing it provided was a default level of respect, a logo on my resume that validated my competence. But I didn't learn anything novel there. In fact, most of what I do now is motivated by the things I disliked at Microsoft; my approach is the exact opposite of how Big Tech operates.
When you're 21, 22, or 23, with no responsibilities or obligations, you have years you can never get back. If you spend them just logging in for a 9-to-5, fixing a ticket, and logging off, you're wasting incredibly important time. We decided, if we're going to do it, why not do it now? We figured we could handle the more responsible thing later. Throwing ourselves into the ether and seeing what worked was the right choice.
Then, I got deported from the United States and banned for five years. It wasn't for anything illegal. I quit Microsoft, moved back to Vancouver, and planned to build things in my bedroom for a year. One weekend, I tried to visit my old roommates in Seattle. The border agents were suspicious, questioning where I was going and what I did. They turned me around at the border, suspecting I intended to stay longer than the two days I claimed. That came with a five-year ban. That was the moment I realized I had to build a company because I had no fallback plan. I was terrified.
I remember driving back from the border, almost in shock. It took a couple of days to calm down, but after that, it was pure focus. I just worked as hard as I could for the next six months.
Rule 2: Prove Yourself Wrong
I tried everything, building anything that seemed remotely valuable. I built video game moderation software in VR, general trust and safety tooling, bot detection software for web traffic, and an anti-cam platform. I would build a ton of stuff, create an MVP, and then try to sell it to see if there was market interest. I did this over and over, experimenting with a new idea almost every week and quickly learning it was a bad one.
The more I did it, the more I got used to proving myself wrong. I learned the counterintuitive fact that in startups, you're actually chasing proving yourself wrong. That's the best thing that can happen because it saves you weeks or months of time. In the beginning, I was building ideas for months, hoping someone would prove me right. That's the opposite of what you should be doing. You should be hunting for someone to tell you why your idea won't work. If you can't find a strong reason it won't work, then you might have a tangible idea to pursue. If I could do it all over again, I would hunt for that strong refutation.
Building as much as you can provides the most information. I don't think I would have ended up building Gum Loop if I hadn't failed ten times before. Talking to users is a privilege you have to earn. In the beginning, you have no users and you're begging people to try your product, which isn't fun but is a necessary phase. Putting users at the forefront of everything, even when they say things you don't want to hear—like your product sucks or isn't solving their problem—is where you get the most valuable feedback.
When I started working on the current version of Gum Loop, I stayed at the office until midnight because I was so excited. I couldn't get enough of it. When you find that thing that makes you want to work until midnight, every day becomes easier, and you build a crazy momentum.
The idea started with AutoGPT, a popular open-source agent framework that took the world by storm. It was the first time it felt like AI could solve a problem on its own. I saw it on Twitter, tried it, and it seemed amazing. I joined the Discord server, which was growing exponentially, and saw countless people asking basic questions: "What is GitHub?" "How do I use a terminal?" "What is a dependency?" I realized I could solve a problem for them by building a simple UI.
My initial thought was that it would be a good way to learn how to build a front end; I didn't think it would become anything special. I called it AgentHub, and whenever someone in the Discord asked for help with their local setup, I'd send them a link. The idea evolved into a "GitHub for agents." But that idea crumbled when I realized the agents themselves weren't useful. That was the aha moment. People wanted to use my platform but were frustrated because the agents were so unreliable.
So, I gave them what they were secretly asking for: reliability and predictability. Their use cases were simple, so I built a framework to automate steps sequentially. That naturally grew into this crazy automation platform. It was originally for developers, but the audience that went crazy for it was non-technical—business admins, ops people, and HR professionals excited about AI automating their work. Realizing 80% of the audience was non-technical was when I knew I needed to build something approachable and fun for them, without the frustrating complexity.
We got into Y Combinator five months before the batch started and kept the product free. During the first week of the batch, we turned on pricing at $20 a month, figuring we couldn't charge more than ChatGPT. Our first paying customer, a guy named Kai, paid $20. We freaked out; seeing that Stripe notification was the greatest moment ever. He's still a user today.
Rule 3: Real Networks Aren't Built at Cocktail Parties
I was stuck in Canada during all of YC. I had all the pressure of needing to build something amazing and live up to YC's expectations, but I had none of the distractions. I was in a small studio apartment in Vancouver, just coding as fast as I could.
We learned that staying focused and avoiding distractions like networking events and tech parties is powerful. The people building something amazing aren't at those events. No one is really networking if you're on to something. I've kept that attitude. I don't normally go to events, and my co-founder almost never does. Most people haven't even met him because he's just working all the time.
The biggest takeaway is that if you stay focused and talk to users, your network will emerge naturally. Many people think you have to go out and network to find investors and convince them of your idea. But if you build something exceptional, they will come to you. You just have to show them that you'll succeed without them. The biggest realization during YC was that it's not that complicated: just build something great. A network is not made at a cocktail party.
Rule 4: Great Products Aren't Built in One Click
There's an anti-pattern with AI where you go too far and become the person with 50 AI agents running your company. I think that's the wrong approach; you're just making a slot machine, creating "slop." It's important to know what you should and shouldn't use AI for. The key is to keep the human touch in the important parts and automate the repetitive things. The best users are super AI-enabled, not replacing their entire job with AI.
But it's a slippery slope. Everyone on Twitter is selling the dream. These "course bros," as I call them, promise you can make $30,000 this weekend if you just copy their workflow. I guarantee that's never worked. They're shilling a vision of productivity without offering anything novel. If there were a magic solution to make $30,000 in a weekend, they wouldn't be giving it away on Twitter.
The value comes from applying AI to something you understand deeply. I see the same patterns emerge whenever there's a hype bubble, whether with crypto, NFTs, or AI. There's a vulnerable section of the community that's easily persuaded that something will save them. You can sell hope really easily, and many people online take advantage of that with fictitious content. They sell a vision of skipping the hard work and shortcutting to the value, which will never work.
I only automate the things I really understand. If you automate something you don't understand, it's just a slot machine. Using AI to do something you don't understand at all creates uncertainty. For example, if you're using AI to code but don't know how to code, you're essentially making malware. It will come back to bite you. Vibe coding can only go so far.
I apply AI to speed myself up. I take the things I do understand and do them way faster so I can learn more and grow. I'm never trying to shortcut understanding or have AI replace me.
It's possible that the last generation of great engineers has been born—an era where you needed to understand what was going on before being accelerated by AI. Now, people can skip the understanding part. There will be a much smaller community of people who use AI as a learning tool to grasp the fundamentals. But it's a slippery slope because it's so easy to not want to understand why something works. If you have the determination to pause, understand the problem, and have AI teach you, you'll become exceptional even faster. The average person will just fall to the slop.
Rule 5: Hiring is Like Dating
Almost everyone we've hired has come through our network. A lot of them were actually customers. Our customer from Instacart quit his job to join us. Our customer from Webflow did the same, as did someone from Shopify. They already had conviction. They loved the platform so much they decided to drop everything and contribute to the mission.
It's a fun way to hire because the only thing you really have as a startup is optimism. You need to be excited to come to work every day. These people are already bought in; they use the tool every day and see the vision. The transition is quick.
Hiring is like dating. You have to be the person someone would want to date. You can't just beg someone to join your company. You have to build something amazing and have traction to make the best people on earth want to join you. But it all starts with just doing it. My co-founder only joined because I had an early, working version of the product that he got excited about.
At Gum Loop, everyone is equally excited about the mission. There are no set hours; people stay late because they love what they're doing. A big filter for my hiring process is: "Do I actually want to spend all my time with this person?" This has compounded over time, and now we have a group of ambitious, intelligent, and fun people to work with. The more people you hire like that, the more the momentum builds.
There are a million reasons not to build any startup. It's easy to hear an idea and ask, "What's the moat? Why won't a bigger company just crush you?" The people who obsess over those questions will never build anything. They end up as pawns in a big player's game. But if you take a risk and try to prove people wrong, you'll end up somewhere that makes people ask, "How did you get there?" The answer is just: you tried, and it worked. And when it didn't work, you tried again.
The biggest quality that makes a founder is thinking they can do it. You will never start a company if you don't think you're the person to do it. It takes this blind confidence, and it's exciting to see just how much someone can do when they believe they can.