Recent clients
The barrier isn't that people can't. The barrier is trust: almost no one is afraid of not knowing — they're afraid of being responsible for an agent's mistake. I remove that barrier. I bring AI into the team and the processes so that within two or three weeks people start delegating real work to agents — and getting real results.
Understanding → trust → delegation.
You can start with a four-hour workshop for leaders, or take a program for the whole team. We fit the format to the goal and the people's level — not the other way around. I run it in Russian and English.
One dense session for a team of executives or marketing leads. A map of what's possible without the magic, live demos on your real work, and a sandbox where everyone delegates their own task to an agent. You leave with a prioritized map of where to start. Works both as a quick way in and as a diagnostic before a bigger program.
Live modules (not recordings), office hours after each one, 1-on-1s on request, homework on the real tasks of your role. Business and engineers go through it together — so they speak one language afterward. The format is already proven on teams from five people to twenty-plus, at any spread of levels.
Targeted tracks for specific functions — development, art, marketing, operations — and ongoing support for the team afterward. We design them after the first results, around real tasks, not hypotheses.
The program is built around your team, your stack, and your tasks. No two are ever the same.
Team questionnaires, your stack, the real tasks, and exactly where things get stuck. The program is assembled from that — around you, not off a standard plan.
Modules live, office hours after each, 1-on-1s on request, homework on real work. Everyone runs their own task through an agent and gets a first win on their own process.
Recordings, a summary of every module, prompt templates by role, personal starter kits, playbooks, and an internal training record for the EU AI Act. The program ends — the materials stay.
The first format works as a diagnostic: it shows where to go next — targeted tracks, ongoing support, a wider program. You make the call, on facts now, not promises.
Engineers, managers, and executives all understand what to delegate to an agent, how to check the result, and how to refine it. The gap between the people who get it and the people who are afraid disappears.
Not abstract tutorial examples but real work tasks run through an agent during the program. People leave feeling "I've already done this," not "I should try that sometime."
Module summaries, homework, prompt templates, personal kits, playbooks on which model fits which task. It all stays inside the company and keeps working after the program.
Written-up feature requests and repeatable workflows instead of hand-improvising every time. The base that real process automation grows from.
A prioritized opportunity map (impact × feasibility) and an honest read on your team's level — the basis for a go / no-go on a wider program.
An internal training record under Art. 4 — a finished document in the package, not just another piece of paper. More below.
As of August 2, 2026, the EU AI Act requires companies to be able to show that staff working with AI are trained. You don't need a certificate — you need an internal training record (Art. 4). Most companies have no such record at all, and there's no official certification for AI trainers in the EU. You leave with a finished document that covers the requirement — not just another piece of paper.
Below are anonymized cases. The industries differ on purpose: the method is the same, and it works in gamedev, in jewelry, and in medtech alike.
A team with a wide spread of levels, from beginners to advanced. A compressed two-week core course doubled as a diagnostic — its results shaped targeted tracks for development, art, and game design. The work ran in a sandbox, on the teams' real tasks.
Here agents were already deployed across eleven roles in the company and integrated with the CRM and marketplaces — the training ran not in a sandbox but in production. A separate session for executives: how to delegate a process, not a text. The client called the program "a turning point for the company" and noted: "by the third week people's eyes look different."
A Spanish company, a four-hour workshop for a team of marketing leads, in English. Live demos on the company's real work — competitive intelligence, lead-quality analysis, localization. It ended with a map of pilots, prioritized by impact and feasibility. People left knowing what to do with it next.
A team of business roles and engineers, three live modules over three weeks. The main result, in the organizer's words, was shared context: "engineers and managers speak one language." The diagnosis held here too: people weren't using AI not because they couldn't, but because they were afraid of being responsible for a mistake.
"A turning point for the company" — that's how one client described the result. Not "we tried AI," but a shift in how the team works.
We start with a short conversation: where the team is stuck, what you'd like to delegate to agents, which format fits. Then a workshop or a program built for your situation. No off-the-shelf packages, no presentations about the future.