☀️
Message on WhatsApp

Claude Code
for
product managers

Enter

What is Claude Code?

Claude Code is a universal AI agent powered by Anthropic's top-tier Claude Sonnet model. It reads and writes files on your computer or in the cloud, executes commands, writes and fixes code, processes data, creates presentations, graphics, video and audio, browses the web, and integrates with other AI models — essentially anything you can do on a computer.

Beyond developers, millions of people from diverse fields use it: from product managers to lawyers, from analysts to marketers, from designers to project managers.

How people use Claude
Connecting…
7 weeks of collaborative exploration and hands-on practice with one of the most powerful AI agents. ⏺ Working with documents: organizing, creating, processing, merging, converting ⏺ Packaging repeating workflows into Claude Skills ⏺ Creating interactive AI features in Claude Artifacts ⏺ Building functional app prototypes ⏺ Building AI automations, websites, web apps, Telegram bots, experimental applications and interfaces. ⏺ Using other AI models to optimize costs ⏺ Connecting external APIs and . ⏺ Deploying to cloud services (Vercel, Netlify, Railway) Never a dull moment.

What participants say

Feedback from Lab Cohort 02 (February 2026). Most participants are not developers.

"The most important thing that changed for me is the speed of going from thought to action. From idea to something ready — minimal 10 minutes of dictation through Whisper, and it's simply genius."

— Luka, Entrepreneur

"You always have this confidence at hand — knowing you can recall any context and ask very specific questions. It gives an incredible feeling of increased productivity and self-confidence. Claude complemented this side of my brain, my personality, and it works amazingly."

— Dmitry, Product Manager

"There's simply no boundary anymore of what you can't do."

— Mikhail

"The huge positive that Claude Code brought back into my life is this state of flow, where you're constantly building something."

— Alexander, Investment Analyst

"What used to scare me and felt like a barrier — has largely stopped being a barrier. With this tool, you no longer have to be afraid."

— Karina

Projects built by non-developers:

  • LinkedIn network search agent with intro path finding
  • Complete business sale package: pitch deck, financial models, due diligence
  • Telegram bot with auto-generated analysis dashboards
  • Bank tender competitive analysis (presented to client)
  • Daily task automation and bureaucracy handling

"The word 'flow' came up from participants independently. People describe working with agents not as using a tool, but as a state. The boundaries between 'can' and 'can't' shifted — not toward 'now I can code,' but toward 'now I can solve problems I wouldn't have even started before.'"

Who it's for

My labs attract entrepreneurs, researchers, developers, generalists, and specialists. People ready to invest time, attention, and effort into building the skills of the future.

Most participants are not developers, and many have no IT background (but basic computer literacy — browser, text editors, search — is expected).

  • For entrepreneurs. If you're looking for ways to optimize your business, discover new niches, and gain structured hands-on experience with generative AI and agents.
  • For multipotentialites. If you've always felt too big for one profession, you develop multiple professional directions, run blogs and newsletters on diverse topics, and picking just one thing is simply impossible.
  • For professionals. You're a product manager, designer, art director, researcher, CPO, or CTO, and want to understand through practice what AI agents can do today — and imagine how they'll work in 5 years, and what your profession might look like then.
  • For developers. To get an overview of technologies and platforms accelerating development. To build the habit of using agents where it makes sense. To understand how Software 3.0 works and how it differs from traditional development.
  • For those ready to act. You want your key projects to deliver tangible results: revenue, connections, visibility, social capital — and you're ready to move from half-baked ideas to detailed specs, process descriptions, software prototypes, and hypothesis testing.

My labs are for those who value a rational, evidence-based approach. We follow research and best practices from LLM developers and proven approaches to working with AI agents. Yes, this is actively studied and published in scientific papers.

Your project

In the labs I recommend a practical approach, best realized through your own project. For some, a project might be a business process prototype, an automation, or a full-fledged AI agent. Others might focus on research: exploring tech stacks, use cases, ideas, constraints, and resources.

A large number of case studies and real examples of solving everyday tasks will help you understand how AI agents can be useful in your business, personal life, and interests.

How the lab works

All lab sessions are built around practice. All presentations (made in Claude Code!) are accompanied by live demos, and participants are encouraged to work in Claude Code alongside us.

Live sessions involve active participation. Technical demos as standalone short videos are published between sessions. During live meetings you can use agents and assistants in real time, build prototypes, write and run code, and use data analysis tools — but we recommend setting aside additional time for deep work.

Gleb Kalinin

Creator and host of labs on artificial intelligence and knowledge management, AI product developer and experimental interface builder.

"I've worked 20 years in IT, and most of that time I've been building things with my hands. I'm a multipotentialite — I find it impossible to do just one thing, I have multiple professions, and new combinations keep emerging.

AI agents will let us leverage all existing technologies — and create new ones on the fly. Since ChatGPT launched, I've been using it and other models to build my own tools: writing code, automations, experimenting with interfaces — I share my work on the Tool Building Ape Telegram channel and on GitHub.

I've been running labs on AI and using agents for creating economically meaningful work since 2022. Over 350 people have participated in my labs on knowledge management and AI agents.

Previous roles: CPO Ozon.travel, Chief of Content Ostrovok, Tsentsiper, and others.

Gleb Kalinin

YouTube channel

Location: Berlin, Germany

LinkedIn: glebkalinin

What the lab covers

Hard skills:

  • Working with modern LLMs and AI agents: Claude Code, Claude Cowork, Codex, Cursor
  • Prompting reasoning models and agents
  • Context engineering and data management
  • Vibe coding and building full prototypes of AI applications with AI agents
  • Translating ideas and projects into the language of specifications

Soft skills:

  • How to deal with FOMO and the endless stream of information,
  • how to build tolerance for errors (which are inevitable),
  • how to plan parallel work with agents and people.

Expanding perspective: How to live in a world with reasoning AI agents?

In the workshops we don't just study technology — we ask big questions about the future. What will the internet and communication look like? What will trust mean when millions of autonomous agents are ceaselessly mining the web for opportunities and vulnerabilities?

What does it mean to automate 80% of office work?

What are the irreplaceable human competencies? What is the role of the specialist and the generalist in the future?

/tools toolkit

Terminal

Claude Code runs in the command line, and we'll cover the basics of terminal work. Mac/Windows Terminal, iTerm, Warp, and more.

Claude Code

One of the most popular AI agents, the main character of the lab.

Claude Cowork

Claude's collaborative workspace where Claude Code runs inside the official desktop app with a visual interface.

Codex

OpenAI's coding agent, a Claude Code competitor. Runs the latest GPT-5-Codex model, and also works in the cloud as a service.

Cursor

The most popular AI agent in IDE form. Can work alongside Claude Code.

GitHub

Version control system that helps save and track code changes, and publish them to the web.

Zed

Code editor and markdown editor with built-in Claude Code and Codex support.

Vercel

+ Vercel AI SDK. Platform and toolkit for developing and running AI applications.

Obsidian

A local text editor that works great with Claude Code.

…and more

The list of tools can be endlessly long and specific: new tools appear constantly

Tools change. The habit of thinking, articulating, and building stays

In the lab we don't just dive into learning tools — they'll quickly become outdated. To live in a world with agents, hire them and get results, we need to learn to think systematically.

We'll learn to formulate tasks so that technology works for you. Describing tasks, detailing context, formulating requirements, thinking in structures and algorithms. A specification written in natural language is what turns an idea into a real product.

/schedule

schedule

Session structure

Starting April 7, 2026
online
meetings on Zoom, coordination on WhatsApp

7 weeks (including a break week) — enough time to build familiarity and gain practice working with agents, understand what they can and can't do, and automate routine tasks with Claude Code and Claude Skills.

Every week we review case studies, practice together, answer questions, and discuss AI tools that help with learning and focus

Mondays 18:00 CET / 12:00 ET / 09:00 PT. Theory + practice sessions, the main meetings. We review case studies and their implementation with Claude Code.

Thursdays 18:00 CET / 12:00 ET / 09:00 PT. 2-hour live coding sessions. We solve real problems together with Claude Code.

No session the week of April 21-25 (break week).

Program

Week 1

Mon 04.07 / Thu 04.10

Intro to Agents

  • What are AI agents
  • Installing Claude Code
  • Core commands and modes
  • First experiments

Week 2

Mon 04.14 / Thu 04.17

Prompting & Context

  • Software 3.0
  • Prompt engineering for agents
  • Context management
  • Context rot and solutions

Break Week

Apr 21 — Apr 25

No sessions

Catch up on materials, work on your project, rest.

Week 3

Mon 04.28 / Thu 05.01

Subagents & Research

  • Agent architecture
  • Parallel workflows
  • Research pipeline
  • Working with transcripts

Week 4

Mon 05.05 / Thu 05.08

MCP, Skills, Hooks

  • Model Context Protocol
  • Creating skills
  • Automation with hooks
  • Integrations and plugins

Week 5

Mon 05.12 / Thu 05.15

Agent SDK & Deployment

  • Claude Agent SDK and packaging custom agents
  • Cloud deployment
  • Best practices

Week 6

Mon 05.19 / Thu 05.22

Practice & Presentations

  • Mon: Q&A and practice
  • Thu: Project presentations
  • Process review
  • Feedback

/pricing

pricing

Participant

Standard option for self-directed learners

€400
  • Access to recordings of all sessions and guides on using Claude Code for data processing, content generation, working with MCP, and deploying apps
  • Access to Custom GPT that helps formulate problems and projects, test ideas, explore technologies
  • Access to NotebookLM with documentation, technologies, and approaches
  • Access to materials: video tutorials, presentations, prompts, project examples, external resources and guides
  • Technical support chat
Pay by card

Team

For groups up to 7 people

€2500
  • Dedicated group chat
  • 4 separate 60-minute consultations / live coding sessions for your team
  • Help with task formulation, tech selection, AI research
  • Access to materials: video tutorials, presentations, links
  • Teams that participated in my labs recoup their investment within 1 month
Message Gleb on WhatsApp

Payment available via Stripe (card), PayPal, IBAN bank transfer (German bank), and USDT / other cryptocurrencies.

For companies, we can issue an invoice from a European sole proprietor (selbständig) with IBAN payment. Contact Gleb on WhatsApp.

Frequently Asked Questions

Can my company pay for this?

+

Yes, we can issue a formal invoice from a European sole proprietor (selbständig, registered in Germany). Please send your company details to Gleb on WhatsApp.

What are the computer requirements?

+

Claude Code supports Windows, macOS, and Linux. You'll need administrative access to your computer for installation. If you plan to use Claude Code on a work computer, make sure installation is permitted.

Are development skills required?

+

The lab is suitable for both people with and without development experience. We thoroughly explain how to use the terminal and other tools, and for most tasks we show how to use no-code tools (like Claude Artifacts, Custom GPTs, NotebookLM, and advanced LLMs). However, we expect participants to be comfortable using a computer (web search, basic Google skills, office software, basic ChatGPT experience).

How are sessions conducted?

+

Sessions are online via live screen-sharing meetings on Zoom. All materials and recordings are available to participants. Theory and hands-on assignments are separate — short theoretical and practical tasks are available after each session.

What subscriptions are needed?

+

You'll need a paid Claude subscription (Pro, Max, or Team).

How much time should I dedicate?

+

We recommend 3-5 hours per week for studying materials and completing practical assignments. However, you choose your own pace.

Are individual consultations available?

+

Yes, the "Individual" tier includes personal consultations for discussing your specific tasks and projects.

What tools will be covered?

+

The program covers modern AI agent tools including Claude Code, Claude Cowork, OpenAI Codex, Cursor, and other relevant technologies.

Will there be hands-on assignments?

+

Yes, the course includes practical assignments to reinforce agent skills and build your own solutions based on the technologies covered.

Can I access materials after the course ends?

+

Yes, all participants get lifetime access to course materials — you'll receive a full archive of markdown files with transcripts, tool collections, and practical assignments. Videos are hosted on Fathom and YouTube via private links, preserved in the markdown files with no expiration.

Is there a money-back guarantee?

+

Yes, within the first week you can withdraw and receive a full refund, no questions asked.

Glossary

Key terms and concepts covered in the lab

Agents

AI Agent — an LLM with tool access, operating in a loop

Subagent — an agent invoked by another agent for parallel work

Orchestrator — the main agent coordinating subagent work

Task Tool — tool for launching up to 10 parallel subagents

Agent SDK — library for building production agents

Models

LLM — Large Language Model

Haiku — fast and cheap model for simple tasks

Sonnet — main model for daily work

Opus — top model for complex tasks

Context & Prompting

Token — unit of text (~4 characters), the basic LLM element

Context Window — agent's working memory (up to 200K tokens)

Context Rot — quality degradation as context fills up

Lost in the Middle — information loss in the middle of context

Prompt — instruction for the model

Meta-prompting — when an LLM creates and improves prompts

Software 3.0 — programming through natural language

Claude Code Commands

/compact — context compression

/clear — context clearing

/resume — continue a session

/plan — planning mode without code execution

/agents — subagent management

/memory — view and edit memory

/cost — current session cost

AI Capabilities

Extended Thinking — deep reasoning mode

Streaming — real-time output streaming

Checkpointing — automatic file change tracking

Session Management — state preservation between sessions

Tools

MCP — Model Context Protocol, standard for connecting services

Skills — modular skills in .claude/skills/

Hooks — automatic actions triggered by events

CLAUDE.md — project long-term memory file

YOLO mode — mode without confirmations

Ralph — infinite autonomous coding loop: while :; do cat PROMPT.md | claude-code ; done

Ecosystem

Plugins — extensions with skills, commands, MCP servers

Marketplace — catalog of plugins and extensions

Slash Commands — custom /name commands

Development

Vibe Coding — creating code through dialogue

CLI — Command Line Interface, terminal

Deploy — publishing to the cloud

Vercel/Netlify — deployment platforms