Я строил команду годами. Теперь ОДИН процесс заменяет людей

2017–2022: Trying to build teams before systems

• Started multiple businesses, including:
  • discount systems
  • hookah outsourcing
  • advertising agency
  • shoe dry cleaning
  • cleaning
• Hired in different ways, beginning with:
  • friends on revenue share
  • later shifting to salaried employees
• For ~6 years, processes were “in my head”
  • Onboarding required personally teaching each role from scratch
• Result:
  • hiring didn’t scale
  • delegation broke down because tasks weren’t documented
  • the “problem” was blamed on people (lazy/unreliable), not on process clarity
• In 2022:
  • he left the projects
  • earned his first net 1 million rubles
  • then rebuilt the team model

Core operating shift (strategy/leadership): process-based execution over role-based reliance

• Key belief: people won’t fix a murky process
  • they often make chaos worse
• Approach:
  • document every process (knowledge base + video tutorials/courses)
  • delegate execution to:
    • AI systems (neural networks)
    • freelancers
    • contractors
    • new hires trained via recorded materials
• Management principle:
  • don’t choose between hire vs automate
  • choose when you’ve described the process well enough to compare execution options

Playbooks / frameworks explicitly used

Process-to-automation prompt structure (5 lines / criteria)

1. Input: what you provide (files, context, links, data)
2. Role: what the worker/agent should do (collect/check/translate/etc.)
3. Prohibitions: what is forbidden (sources, wording, data to not touch)
4. Output artifact: one concrete file/table/message (not “a report”)
5. Verification: how results are checked (e.g., weekly report with what/how/results)

Outcome-as-algorithm logic

• If the outcome is clear and can be expressed as:
  • an algorithm
  • a skill
  • an agent
• …then it can be executed reliably (by AI or by trained people).

Delegation readiness rule

• If you can’t list the inputs and verification rules, then the task isn’t ready for delegation/automation.

Concrete business examples / case studies

Marketplace cards → Ozon (independence + capital accumulation)

• Pivoted to design + sales to become independent and accumulate capital
• Reported early earning ramp:
  • “several hundred thousand rubles/month” from marketplace cards
• July 30, 2023:
  • recorded 1 million rubles profit net “without any team”
• After that, launched Ozon activity
  • (turnkey store/presence implied via marketplace execution)

Marketplace sales management agency (delegated; founder not involved)

• The agency handles:
  • selecting products for clients
  • setting up advertising
  • launching turnkey marketplace stores
  • client communication and operations
• Stated current model:
  • he is not involved at all
  • no meetings/training
  • doesn’t need details like internal staffing (e.g., number of product selectors)
• Profit metric:
  • about ~0.5 million rubles, on average across the year
  • varies by month (“depends on the month, about half a million”)
• Training system:
  • internal role-specific lessons
  • he claims ~30+ lesson “lessons” are recorded
  • new hires can become operational in about a month
• Hiring strategy example (manager)
  • instead of hiring an experienced manager, he hired a manager as co-owner with no experience
  • trained via parallel lesson delivery
  • first month: “so-so,” after a month: stable and more efficient operations

AI + neural networks for execution automation

• Neural networks used for:
  • routine automation (e.g., salary calculation/analytics)
  • creating products
  • small operations like YouTube cover generation “in their style”
• Also used for YouTube operations (at least one channel)

YouTube production scaled with minimal headcount

• Runs 3 YouTube channels
• Posting cadence:
  • ~8 videos/week
  • long videos of ~30–40 minutes
• Team:
  • 5 production staff
  • 3–4 editors (freelance/part-time)
• Claim:
  • editors can be outsourced because editing processes are fully documented
  • templates + a beginner-friendly editing course

Telegram example (organizational scale argument)

• Mentions Telegram reportedly having ~30 engineers at ~1 billion users
• Argument:
  • employee count isn’t the driver of product quality
  • coordination overhead slows large organizations
• Cites Pavel Durov:
  • “How big is your HR department?”
  • interprets HR as effectively zero
  • implies decentralized hiring/engineering

Historical examples used to support “process > headcount”

• Instagram sale to Facebook for ~$1B with ~30 employees
• Journey/Midjourney-style comparison:
  • small teams reaching large revenues
  • example given: ~10 people reaching ~200M annually
  • later ~100 people at ~500M
• Founder funding philosophy:
  • big money turns researchers into quarterly-review managers

Key metrics / KPIs and numbers mentioned

Net profit milestones

• 2022: first net 1 million rubles after going solo from prior projects
• July 30, 2023: ~1 million rubles net profit “without any team”

Income levels

• Marketplace cards stage: several hundred thousand rubles/month
• Design solo ceiling: around 200,000 rubles/month
• Freelance delegation model: up to ~300,000 rubles
  • splitting orders with an assistant at 50/50

Agency profitability

• Marketplace agency profit: ~0.5 million rubles average across the year
  • varies by month

YouTube cadence & resourcing

• Posting: ~8 videos/week across 3 channels
• Team: 5 production staff + 3–4 editors
• Video length: 30–40 minutes

Training/implementation timelines

• New hires using the process/course become operational in ~1 month (per manager training example)

Actionable recommendations distilled

• Stop blaming hiring quality; fix task definition
  • if delegation fails, the fix is process documentation, not “better people”
• Create a knowledge base + video tutorial system per role
  • record repeatable lessons for each function (selectors, analytics, client comms, finance/spreadsheets, etc.)
• Make tasks “prompt-ready”
  • use the 5-line structure (Input/Role/Prohibitions/Output artifact/Verification)
  • so AI or contractors can execute consistently
• Delegate only repeatable work
  • if tasks are murky, automating/hiring without context scales chaos
• Hire after building process clarity
  • hire if you can’t execute yet—but ensure training materials exist so execution doesn’t rely on tacit knowledge
• Use AI where “style + outcome” can be specified
  • document standards so AI can produce consistent outputs (e.g., creative covers) and run routine analytics

Presenter / sources

• Presenter: Roma Wright (author of the video/channel)
• Referenced sources/citations:
  • Pavel Durov (Telegram/HR/scaling quotes)
  • Lex Fridman (mentioned as the interview context for the Durov conversation)