A small organism with good tools.

I started coding from scratch in October 2024. The first thing I built was the personal dashboard that's still on zacc.ai today — what later became Drako iOS. First commit landed in June 2025. Everything since is downstream of that one project.

Before software: architecture in high school, mechanical engineering in college, a data-science internship after graduation, then years training clients at Equinox, designing a restaurant in NYC, running the NYC Marathon in 2025, training MMA, coaching wrestling at Brooklyn Tech. Software is the latest thing. By day I work at Yuma AI — a subnet inside the Bittensor network.

Drako is both my alias and the umbrella over the work. Drako DevOS is the operating layer everything else runs on top of; the rest are verticals proving out what one person plus a swarm of agents can build on the same substrate. Seven projects, one operator.

"A swarm of agents under one operator beats a team of ten. A one-person studio with the right operating layer can ship what used to take a startup."

I have ADHD, and what I learned the hard way is that ADHD is mostly a routing problem. Attention isn't missing; it's mis-allocated. The cost of mis-allocation is paid in context-switching tax, in dropped threads, in arriving at the desk with seventeen open tabs and the wrong one on top. Most productivity systems try to fix the human. The tools I build try to fix the routing.

Two great collaborators beat ten mediocre ones — that's a rule I learned in design and it generalized cleanly. The same logic now applies to the AI agents I work alongside daily. A small set of capable, specialized agents beats a large pool of general ones; one operating layer that holds them coherent beats a dozen consoles in a dozen tabs. That's the load-bearing claim under everything Drako is.

Taste came from interior design. I look at software the way I used to look at rooms: what does the eye land on first, what's holding the weight, what's load-bearing and what's decorative, where does someone trip when they walk through, where do they want to sit down.

The 2010s were about platforms — single companies, large teams, network effects. The 2020s' first half was about scaling the platform pattern with more capital. The second half is going to be about small studios: one or two operators with a dense set of capable AI collaborators, building things that previously required a team of fifteen.

That's not a prediction so much as it's the only thing that explains what's happened to my own week-over-week throughput since I started in October 2024. The leverage isn't theoretical anymore. It's just unevenly distributed.

Drako is my attempt to build the operating layer that makes that distribution narrower. If you can give one founder a substrate good enough that a small set of AI agents feels like one organism — what gets built next looks more like a small studio than a startup. The studios will outnumber the startups by 2030.