For the last six months, I've been in the trenches.
I thought I knew what it would be like to start a company. I had taken what I thought were decent swings at this game before. I think I was really just getting my feet wet.
I call these 'naïve ventures'. They were projects I started, usually alongside others, that told a story about the world that I wanted to contribute to. But in reality, they were nothing that I would have ever quit my day job for.
So when I did quit my day job to start a company, it felt like jumping into arctic waters with no wetsuit after spending the last hour in the hot tub.
It's not like my previous ventures didn't teach me valuable lessons. I don't think I'd have made it this far without those experiences.
My first venture, talentDAO, received almost $100k in grant funding over its life to help prove that that DAOs could be productive organizational units of society.
We didn't survive. We didn't ship product fast enough. Our efforts were spread thin across a concentrated network of async part-time contributors who were all trying to do their own thing. We had a lot of great ideas, but we didn't have the right people in the right roles to execute on them.
The problem with DAOs is that they aren't companies. Companies, for good reasons, are not democracies. In fact, almost all successful companies that exist in the world today run like dictatorships. Ironically, while it was my background in industrial psychology that originally attracted me to DAOs, I was blind to what just about anyone who's spent five minutes working in Corporate America knows intimately.
Still, building talentDAO was the first time I felt like I was meaningfully contributing towards something bigger than myself. I didn't feel this at SpaceX. Partly because I wasn't a rocket engineer. But also because it was a big company. Sure it followed the Elon playbook — meaning, the average role had more impact than at most other companies — but it was still a haven for bureaucracy fueled by politics like anywhere else I had been. And I was still a cog in the machine.
Throughout my time at SpaceX, I spent nights and weekends immersing myself in the Silicon Beach tech scene, building weekend projects with new friends that would never see the light of day. It was fun for a while. But eventually, I started to dread my day job. I was getting more responsibility than I wanted. And I really wanted to build AI software.
Over the years, I got to meet some interesting people. I met Eric, the co-founder of my second venture, Noometic AI, at a weekly AI meetup in Venice.
Eric was a talented mathematician and engineer. When we met, I was doing research on the application of LLMs in the influencer marketing space. He had just written a thesis on using AI to measure influence in social networks, so our product ideas we're already fairly aligned. Those focus areas are what led us to building what I would best describe as Perplexity for finding content creators.
The biggest challenge for us was building the data operations to index and scale a search engine. We built a dataset of millions of creators that took us much too long to curate. Looking back, this was a clear cold start problem that we should have anticipated, but we underestimated the complexity and importance of factors like data enrichment and ranking algorthims. To put the difficulty of the creator search problem into perspective, consider your favorite influencer and try to imagine searching for them on social media without using their name, any associated tags, brands, or going through friends pages. Now that you're thinking of all the ways you can describe an influencer, imagine doing that for millions of creators across multiple platforms.
Ultimately, we took the route of greatly simplifying our algorithm (see BM25 is all you need) to focus on UX. But after six months of trying to balance a 50+ hour/week day job with 20+ hours/week on a startup that was struggling to gain traction, I was extremely burnt out.
It was time for me to take a break. I put my head down at SpaceX for a while. I got engaged, planned a wedding, and coasted for a bit. Life was good, but I couldn't shake the feeling that timelines were accelerating.
The thing they don't tell you about startups is that once you've decided you want to do one, that feeling never goes away.
I knew that I was going to build an AI company. It was the main thing I was passionate about. When people from the community started reaching out for help with thier projects, I knew it was a sign the industry was catching on.
Over the summer of 2024, I ran into one of the investors we pitched while building Noometic AI. He was doin his own startup again after years on the venture captial side. He was "eating his own dog food", as he liked to put it. I respected that, so I took my first AI engineering contract with him.
That company was called WiseRx. For a few months, I worked on building out a serverless vision system that could turn images of prescription pill bottles to structured data, which could integrate with an at-home care device designed to help seniors take their medication.
For the first couple months I was still at SpaceX. As I worked on other things, the tensions between my two worlds were starting to grow. SpaceX was demanding more from me, but I was happier working on AI.
I managed to convince myself that it was time for me to take the leap. The day after my next vest, I quit my cushy tech job. I was free, but I was terrified.
A couple of years before, I met a dude named Bootoshi. The day we met, I watched him bring an entire NFT project to life with AI characters that could talk and operate in a virtual world. The Boo Kingdom went on to become a thriving community of both humans and AI agents.
When I quit SpaceX, he was one of the first people I reached out to. I'd seen few others in my life match his level of hustle and creativity.
The second person was Rob. We had already built some rapport throughout the years, collaborating in hackathons and hanging out at local meetups. What I loved about Rob wasn't just that he had a decade of experience leading deep tech projects, but that at his core he was a designer, which was clearly reflected in everything he produced.
Somehow, I managed to convince them both to join my team. That was one of my first lucky breaks.
Of all the things I've done right, I think picking good co-founders was among the most consequential.
Together, we founded Agency42, an AI innovation studio that is still just sort of figuring it out.
The lessons I learned through my naïve ventures were invaluable, but the only thing that can prepare you for trench warfare is trench warfare. Don't expect bootcamp to turn you into a super soldier.
Despite just barely getting out of the mud, today I can remain focused on building my vision for the world. One that I'm lucky enough to share with a team of builders that I respect and admire.