The execution layer
Between intelligence
and the real world.
AI is getting smarter every month. Physical operations aren't. Between them sits a layer nobody owns — where a decision turns into throughput, cycle time, and margin. That is what I work on.
AI knows what to do. The pit doesn't care.
Between a model's recommendation and a ton on a ship there is a layer nobody owns: dispatch, sequencing, maintenance windows, the shift supervisor's judgement at 3 a.m. Every failed pilot I've seen died there, not in the model.
I call it the execution layer. It is not a product category yet — it is where the value of industrial AI will actually be captured over the next decade.
I've spent 25 years running heavy industry: coal, steel, precious metals, and freight railcars.
I started underground, at Russia's largest underground coal producer, and later served as deputy CEO of Polymetal, one of the world's major precious-metals miners.
At United Wagon Company I ran business development and international expansion:
{{ vgkStatus }} East Mining Company (EMCO), which operates the Solntsevsky Coal Mine. I joined in 2019, when it was a mid-sized regional producer. Four things came out of it:
Nobody was going to close the gap between the model and the machine from the outside, so we built the layer ourselves — and then spun it out.
I live in Dubai. EU citizen. I write and speak about what it takes to make AI real in the physical world.
The best way to reach me is email.