I wrote my first production code in 2001, on a suite of marketing software. For the next six years I ran web development, digital marketing and the whole e-commerce operation for a distributor - which is where I learned that code only counts when it moves a business number.
In 2008 I founded Data Subsystems and have run it since: 130+ projects for clients in payments, e-commerce, manufacturing, real estate, crypto and media. At some point I noticed my real product isn't code. It's the moment an owner stops worrying about the system - because it runs without me: documented, handed off, on autopilot.
For fifteen of those years the engine behind the speed was ninja teams - small squads of exceptional developers I hired, led and kept for years, running up to ten client engagements in parallel. Clear scope, tight feedback loops, ruthless review, working software every week. Management was never the product; it was how I kept ten clients shipping at once. Those people did not go anywhere: twenty-five years in the industry left me a network of the best I have worked with, and when a job is bigger than one person, a vetted team is up in days.
Today I run the exact same playbook with a different team: AI agents, orchestrated the way I used to run people. Same discipline, same gates, same honesty about scope - but the cycle time dropped from weeks to days, the team never sleeps, and I am back to writing most of the code myself. Nearly ten thousand commits in the last ninety days, and my review is still the gate before anything goes live.
My client relationships are measured in years; the longest ran seventeen. When one client's fashion-retail chapter ended, we didn't part ways - I helped him turn the same operation into a fine-mineral business. I like long games and second lives: my own 2007-era domain died in a Google update and came back sixteen years later as a startup radar.
These days I also write about startups: The Shortov Report - a weekly on where founders and venture money are heading, by email today and on YouTube next. After twenty-five years of building software for them, I have opinions worth publishing - and a radar to check them against.