Road from 7 figure comp to 7 figure ARR
Once in a lifetime opportunity. That's how I described my time at Airbnb. I got to build out the Dev AI team, and I got to do it making more money than I ever thought imaginable. For the last 2+ years I woke up every morning, turned on my work laptop and focused on making 2,500 engineers happier, more productive and generally just able to coexist with a new AI centric company. It was no one thing, and I wasn't one person working on it. I saw it as a chance to help folks transition to a new way of working.
There really wasn't a big reason it was me either. I didn't have an ML background. Best I can come up with is they liked my nonstop desire to make things easier for those around me. I've been described as someone who can get folks to walk off a bridge, which is sort of one of those great power, great responsibility things I need to deal with 😅
So why did I leave it all?
Especially after close to doubling my comp in 2 years? I left to do it for everyone.
I believe the way we build software is fundamentally changing. I wouldn't say it's easier, but it definitely is different. The gap between what a beginner and a more seasoned "AI native" developer can do is a bigger multiplier than any other I've seen in tech. After many conversations and ample support from my friends and family, I decided to help multiply everyone.
I left my dream job for one reason: to take everyone to the frontier of AI development. It's no one thing that gets you there, but I can tell you few backtrack after getting there. That's the road in the title, from 7 figure comp to 7 figure ARR, and we're at the start of it.
Yesterday I published a post about how I got here. Today I want to dive more into what's next.
Coworking with AI
A small team and I are working on https://hive.codes, a platform for humans to work with AI agents. The way I see it now, we now cowork with AI more than we use it, and we (Friendly AI Labs) want to help you with the tools, processes and infra to be happy doing it.
We are all at different parts of the adoption curve. I see 5 levels of AI usage:
- Copy paste from a web chat, conversational only
- Autocomplete in editors and IDEs
- Agents in the terminal that can call tools on their own
- Agents in a web experience, running off your machine
- Orchestration and factories: coordinating many agents at once with loops and graphs
Each level comes with its own needed guardrails. Giving agents tool access means guarding against what the tools can do. Orchestrating agents means having a communication channel between them.
The hard part
The edge of the adoption curve, level 5, is folks with scheduled routines and event triggers that seem to do their work for them. They do spend a lot of work getting their loops, graphs and sandboxes to do what they want, but the payoff is massive, they can walk away from work and have it keep working. Almost anyone can now set up a system where metrics and analytics automatically trigger agents to fix things, or where a Jira label triggers another agent to do something.
The thing is, most don't realize we can all do it now. We don't need crazy funding, GPUs or big teams. Anyone can orchestrate dozens or hundreds of AI agents. Doing it safely, efficiently and without making maintenance your full time job is the hard part. I like hard parts, and I'm making this one my job.
I am going to be your DevAI platform team. https://hive.codes/ is the dev platform you're already using or building, just managed for you. In the coming weeks we'll keep releasing more technical content about how things work and why we think you'll want to work in Hive as well. Join us in a happier way to use AI