RU EN EL
Book a call 30m
AI ADOPTION FOR OWNERS AND EXECUTIVES M-2 : 1

AI Potential.

I bring AI into your business and measure the result in hours and money

My name is Mikhail Zelenin. I help owners and executives adopt AI so that it pays for itself. I free up 20-40 hours a week. I take the routine off your team: reconciling CRM and 1C, standard emails, contract review. I launch processes that run without days off. Not a marathon on neural networks, but concrete processes with numbers before and after.

«Not yet another neural-network course. A working AI loop, live in production, that pays for itself in 4-6 weeks.»

AI architect · 20 years in IT · 4 products built solo

DIAG. A-1
LIVE · 2026
ai-sys arch.
AI POTENTIAL AI CONSULTING · ADOPTION
0 · 1000mm · BRINGING AI INTO BUSINESS
§ 01 / AI STAFF

I bring AI employees into your business

// THE AI-EMPLOYEE ERA

Businesses used to be reshaped around mobile, so the customer could get through from a phone. Now it is around AI employees: agents that work in your loop without days off.

An AI employee does not get sick, does not quit, does not leave for a competitor. No income tax, no insurance contributions, no talent shortage, running around the clock. Against rising taxes and a shortage of people, this is no longer a luxury but a condition for survival: companies with an AI staff get a serious head start.

But you cannot drop an AI employee into a mess. For it to work, it needs access to your data and processes: sales, CRM, 1C, the contracts department. First put that order in place, then launch the agents. That is exactly my job.

DIAG. A-2
ai employees
«I reshape your business for launching virtual employees, agents into your loop.»

That is why I split the work in half

So you do not pay for a rollout where learning is enough, and do not lose months where engineering is needed.

00 · YOU YOURSELF
Part of the consulting

You assemble the first AI employee yourself

I teach you to assemble your first AI helper with your own hands: we set up Codex and work on your real tasks until it runs. After that you repeat it without me. This is not something to pay for as a rollout, it is something to simply learn, and it is part of the consulting.

01 · MY ENGINEERING
My engineering

An AI staff on your systems

I connect AI employees to your rails (1C, CRM, AMO) and move the team onto them. This is no longer a prompt but engineering: I design the loop, hire the agents and put them to work, set up the process.

You will assemble the first AI helper yourself, I will teach you as part of the consulting. A team of agents inside your systems and data is already my engineering.

§ 02 / I BUILD IT MYSELF

I do not consult in theory. I show it on my own products

// ADOPTED IT FIRST ON MYSELF

I build AI products and agents myself, so I speak from practice, not from articles. I see where AI genuinely speeds things up and where it is still just talk. Here is what already works.

Personal operating system (MyOS)

BLD-01

My system on Obsidian and AI agents: tasks, knowledge and decisions in one place. I run all my projects on it, and I set the same thing up for clients.

Psychologist marketplace

BLD-02

The platform matches a client with a psychologist by compatibility. More than 40 practicing specialists work in the system.

The Pochva platform

BLD-03

An IT platform for mental health and personal effectiveness: a neuro-trainer, breathing practices, courses.

Vehicle inspection booking in Cyprus

BLD-04

Booking a MOT through WhatsApp entirely without operators. Inside are six AI agents: they catch requests, find slots, book, and check quality.

// STACK

What I work with every day: Claude Code · Codex · Antigravity · MCP · LangGraph · Ollama

§ 03 / TRUST

What backs it up

// NUMBERS AND DIFFERENTIATION
20+ AI projects in progress
10+ virtual AI employees across projects
8%+ average margin increase
2-3 mo average payback of a rollout

The market is full of prompt engineers, AI bloggers pushing 500 neural networks, and coaches with no real grasp of AI. I work at the intersection and build working systems by hand.

§ 04 / SIGNALS

People usually come to me when

// STATE DIAGNOSTICS
ROOT OF THE PROBLEM

Your peers in the market already count AI in money, while you still do it by hand. A year from now, the ones who adopted AI seriously pull far ahead on speed and margin. Catching up will cost more than starting now.

SIGNAL 01

routine eats 60+ hours a week and someone's salary, even though a single AI agent closes it in an hour

SIGNAL 02

you have the data, but it is scattered: CRM, 1C and Excel live separately, and decisions get made by gut feel

SIGNAL 03

you want to adopt AI, but it is unclear where it pays off and where you would just burn the budget

§ 05 / PRACTICE

What this delivers in practice

// CASES AND VOICES

Names and recognizable details are changed, that is the ethics of the practice. The before-and-after numbers come from real rollouts in progress.

CASE 01 confidential

Bakery chain

AT INTAKE:
Fresh baked goods get written off at the end of the day and eat into the margin. Ingredients are ordered by gut feel: now a shortage, now overproduction. Across locations it is not visible where exactly the money leaks.
INSIGHT:
Not a big AI project, but a single loop: AI gathers sales by location and day, forecasts demand, and suggests how much to bake and how much to order.
OUTCOME:
Write-offs fell from 12% to 5% of revenue across the chain's locations. About 15 hours a week of manual order planning was freed up. The rollout paid for itself in roughly 2 months.
CASE 02 confidential

Support fleet

AT INTAKE:
Requests for vessels land in email and messengers, and dispatchers compile the plan by hand. Hence downtime, scheduling conflicts, and hours spent preparing voyage documents.
INSIGHT:
An AI agent takes in requests, compiles them into a vessel loading plan, and prepares drafts of the voyage documents. The dispatcher only checks and approves.
OUTCOME:
Request handling time dropped from 40 minutes to 5. The dispatching routine shrank by roughly 20 hours a week. Scheduling conflicts became noticeably fewer.
// TRAINING BUSINESSES

AI HQ: an intensive for the owner and the team

Over the intensive, the owner and team assemble their own staff of AI employees and leave with a working HQ, not a notebook of notes. From €5,000.

LEARN MORE →
§ 07 / ACTION

How to start

// START

Introduction call · 30 minutes

If it resonates, I invite you to a 30-minute online introduction call. No obligations, no presentations, no sales. We will find out if there is working chemistry between us and where to dig. If I cannot name a concrete direction within 30 minutes - I write a 1-page review of your question in DM for free. 5 slots per week; nearest ones close 7-10 days in advance.