[nevrai]
· 10 min read

From Managing 150 People to Solo AI Builder

Two years ago I managed 11 product teams and 150 people. Now I work alone and ship more products than that entire team launched per quarter. Here is how it happened.

PUNKT E: Revenue x5.7 During a Market Downturn

PUNKT E is the most popular B2C EV app in its market. 95,000 users, a nationwide charging station network. I joined as Product Lead in 2024.

The EV market was tough: sanctions limited imports, charging infrastructure lagged, users were skeptical. The market dropped 37% during my tenure.

Despite that, we grew revenue from $700K to $4M over 2 years. A 5.7x multiplier. How:

  • Unit economics overhaul. Repriced charging sessions. Introduced dynamic pricing by time of day — peak hours cost more, nights cost less. This balanced network load and boosted margins.

  • NPS from 3.2 to 4.9. The key problems were payment UX and station navigation. Redesigned the payment flow (three taps instead of seven), added real-time station status, push notifications for charging completion. NPS grew organically.

  • AI products. Three internal AI projects: charging session completion prediction (PyTorch, LSTM), first-line support chatbot (LLM), AI market research for strategic decisions.

That last project — AI research — became the prototype for AICPO. I saw that AI could not just automate routine tasks but replace entire analytics departments.

Mafin: Scale and Process

Before PUNKT E, I was Head of Products at Mafin — a Top-3 insurtech. 11 product teams, 150 people in direct and matrix reporting.

What I did:

  • P&L for 4 business lines. Auto insurance, travel insurance, subscription products. Each line with its own economics and team.
  • Expanded the product line from 3 to 16 products. New insurance products, partner integrations, B2B channel.
  • M&A integration. Mafin acquired an insurance broker. I integrated their product and team.

Scale teaches you process: road-mapping, OKRs, discovery, delivery, stakeholder management. But scale also shows how much energy goes to coordination, not creation.

In a 150-person team:

  • 60% of time goes to synchronization (standups, retros, planning, 1-on-1s)
  • 25% goes to documenting and communicating decisions
  • 15% goes to actual product work

I realized I wanted to get back to 100% product work. But without losing the output scale.

Why I Went Solo

Three reasons:

1. AI changed the economics of building. In 2024, Claude Code and similar tools reached a level where one person with AI can do the work of a small team. Not in theory — in practice. I saw this firsthand on the AI projects at PUNKT E.

2. Coordination costs more than execution. In a large team, adding one developer does not linearly speed up development — Brooks’s Law. With AI agents, there are zero coordination costs. An agent does not get sick, does not attend standups, does not need management.

3. Iteration speed. In a corporate environment, the cycle from idea to production is 2-4 sprints (4-8 weeks). With Factory OS, idea to deploy takes hours. That is a fundamentally different learning speed.

Factory OS: What It Enables

Factory OS is my AI agent orchestration system on top of Claude Code. 15 specialized roles: Builder, Strategy, Discovery, Quality, DevOps, Marketing.

What this changes in practice:

  • 9 products in 3 weeks. Not prototypes — deployed products with users.
  • $15/mo infrastructure. Free tiers cover 99% of needs at early stage.
  • Zero management overhead. No standups, retros, planning meetings. Just a task board and memory files.
  • Instant feedback loop. Write task, agent executes, review, deploy. Minutes, not weeks.

What I Lost

Honestly — it is not all upside:

  • No team for brainstorming. AI does not replace a live discussion with smart people. I compensate with communities and mentors.
  • No shared responsibility. Everything is on me: product, code, marketing, support, finances. It is draining.
  • No automatic social capital. In a company, you build a contact network automatically. A solo builder has to do it deliberately.
  • Financial risk. Stable salary replaced with uncertainty. Currently living on savings and freelance consulting.

What is Next

Three tracks:

1. Product. AICPO and InkCloak — two products with monetization potential. Focus on PMF and first paying users.

2. Consulting. AI product audits, building AI pipelines, implementing agent systems. Already done several engagements.

3. Building in public. Sharing what works, what breaks, and what I learn. This blog, the newsletter, and the open-source pieces of Factory OS.

The transition from managing 150 people to solo builder is not a downgrade. It is an upgrade to a new level of leverage. One person + the right AI tools = the output of a small company.

The question is not whether AI will replace product teams. The question is who will be first to use AI for the full cycle of product creation — from research to deploy. I am betting on myself.

For contact: [email protected] | Telegram @nevr_ai