Who is behind gull system and how it started: from engineering competitions to a CAD automation platform of our own.
I started at fifteen, and I never set out to build a product. I just wanted to stop drawing the same thing twice - so I rewrote the geometry into formulas and let the model redraw itself.
For years I built generators that way purely for myself - wings, gears, cams, whole combustion engines. Until, in the middle of one of them, it dawned on me that what I had on my hands was not a tool for one part, but a way to automate the design of practically anything.
That idea is now gull system: a platform that hands the same thing to your engineers. It connects to the models you already have, takes over the routine you do by hand today, and turns weeks of production preparation into minutes. The rest of this page is the story of how I got there.
I was fifteen. Aerodynamics at my school was taught by a man who had the complete calculation of an ultralight in Excel. He wanted it in CAD, but not redrawn: he needed a model that would redraw itself from those numbers.
Excel connects to Inventor - but he could not use CAD, so he went looking for help. He started with the fourth-years, got nowhere, and went down a year, and another, until he reached the first-years. My form teacher put my name forward - I was three weeks into CAD.
He showed me what he needed, and the next day I brought him a cube that redrew itself from numbers in a spreadsheet. Only a cube - but proof that the two programs talk to each other.
Then came the wing generator: a library of NACA profiles, twist, rotated coordinate systems. Two months later the wing redrew itself. I did not know the word configurator then - but it had dawned on me that a model should not be drawn, it should be computed.
Parametric wing profile calculation in Excel
After a year I could take CAD about as far as it goes, and I ran into programming: Python, then C#, HTML, CSS. I built websites for a few firms I knew - it taught me a different craft, an interface that has to make sense to whoever opens it, and work you stand behind. But it was not the thing.
So I went back to CAD and looked for a harder problem. I found it in gears: an involute is a curve you can write down, but the undercut at the root of the tooth is not. So I took it from production - the flank is computed as if a rack cutter rolled over the blank.
All of it in Excel, and all of it written out so that someone could read it after me. The flank grows as a spline through dozens of computed points, and finished gears come out through CAD configurations: spur, helical, even worm. Change the module or the tooth count and seconds later you hold a new model, with the right undercut and profile shift.
The spreadsheet had a ceiling, so I rewrote the whole generator in Python with numpy and built it a web front end: the gear rendered over WebGL, straight in the browser, and could be downloaded as STEP. It was a configurator on the web - I just wasn't calling it that yet.
The involute was joined by a cycloid - gearing with almost perfect rolling contact. I developed it the same way, it was simply easier: I solved the parametric equations of the epicycloid and hypocycloid numerically in the sheet, bisecting the interval in twenty steps.
I can plot the rolling contact step by step, so a mistake shows up before a gear ever reaches the printer. But I still saw a use for it only for myself and my classmates: accurate gears for 3D printing that nobody else would bother making. That it could be a product somebody would pay for never once occurred to me.
Main page of the gear generator in Excel
Then came the year of combustion engines - how they work, the calculations, the air flowing into the cylinders and out. And above all the valvetrain: the cams and their profiles gripped me most, because they decide when a valve opens and when it shuts.
I wanted a model where the intake and exhaust valves open as precisely as possible, so the cams had to be driven. I still only knew how to connect Excel to CAD, so I wrote the cam generator in Excel, with different ramp curves. Out of it came a twin-cylinder with a DOHC belt valvetrain.
The hardest problem was waiting inside the D-cycle engine. It is an unconventional concept: it has two pistons so that it completes four strokes in a single revolution, and the second one is driven by a cam, not by the crankshaft. There is nowhere to copy that profile from.
None exists anywhere, so I computed it point by point - compute, verify, correct - until the whole mechanism ran. I printed it and it ran exactly as it should. It taught me that even a motion no table describes can be captured in parameters.
I still saw no use for the generators beyond myself, so I took on a project nobody believed in: a V6 with a supercharger and a valvetrain running off four chains, like an Audi. When I said it out loud, people laughed - a seventeen-year-old.
I never had the real engine in my hands; I built it from whatever could be found out about it. What came out was a model of more than 1,500 parts, 300 of them screws, and the block alone printed for five and a half days without stopping. The cams and belts were no longer drawn by hand but produced by my generators.
3D-printed DOHC valvetrain model at a competition
3D-printed V6 engine model at the Moderní škola competition
The V6 earned me a nomination to CzechSkills 2026, the Czech championship of professional skills, CAD Designer. Out of more than a hundred nominations thirteen projects were picked and examined in detail - and six of us were left, the best in the country.
The competition ran in Brno in the spring of 2026: three days from morning to evening, classic parts, sheet-metal parts, weldments, assemblies and drawing documentation, every task measured on time and on precision. Three days against the best young design engineers in the country.
I won. Czech national champion in CAD design. I knew exactly what that title was: not a reward for three days in Brno, but for three years of evenings on my own projects - the wing, the gears, the engines nobody had asked me for.
One thing was left: presenting the V6. At the regional authority in Ústí nad Labem we won without the slightest competition, and the contest made it onto television. Then I looked back for the first time and saw that hard work will carry absolutely anything through, including the things nobody believes in.
Then came the question of what I would do for a living - and with it the discovery that CAD can be connected through an API, not only through Excel, and that it is open. That was when I saw where I had been heading.
I connected SolidWorks, and out of it grew the whole platform, the one you set up yourself. Your customer is my generation, used to buying online and at once: I am not selling software as an investment in the future - the future is now.
CzechSkills 2026 - 1st place, CAD Designer
Diplomas and medal from CzechSkills 2026
The gull system application - model configuration