OpenClaw: the self-configuring agent of the future
No Bullsh!t, March 18, 2026, 93 min, 1,096 views. This episode's guest, Maxim Postnikov, is a serial tech entrepreneur and former head of social networks at Mail.ru with two exits, now building in AI and calling himself a vibe founder.
| Published | |
|---|---|
| Length | 93 min |
| Views on YouTube | 1,096 |
| Language of the episode | Russian |
| Where to watch and listen | YouTube, Spotify, Telegram |
Date, length and view count according to data from the episode page on YouTube, as of the page update date shown in the footer.
In short
The conversation opens with a personal story: the path from his first computer in Krasnoyarsk, through six years at Mail.ru, selling two projects, losing himself in the role of startup mentor, and finding his way back to entrepreneurship through AI. From there the talk turns to how working with tools like OpenClaw, Claude Code, and Codex is changing: why the old idea of vibe coding is outdated, how an agent with access to a personal knowledge base in Obsidian differs from an ordinary chatbot, and why handing an agent real permissions and context radically changes the speed of work. They discuss three strata of reaction to AI (the model makers, large businesses with their inertia, and independent entrepreneurs), the risks facing project managers and white-collar workers, a renaissance of solo entrepreneurship, and how to properly hold a dialogue with an agent instead of giving it vague instructions. They close with Maxim's advice to his five-years-younger self and the idea that growing productivity should come with growing ambition.
Key ideas
- Working with modern AI agents (OpenClaw, Claude Code, Codex) is fundamentally different from a regular chat with an LLM: the agent gets long-term access to your personal knowledge base (for example, Obsidian), remembers context, and acts proactively instead of just answering questions.
- The idea of vibe coding as a toy for hobbyists is outdated: modern agentic tools have reached a level where you can build working products and services, not just simple prototypes.
- Society is reacting to AI in three distinct strata: the model makers (OpenAI, Anthropic, xAI), large businesses, whose IT adoption cycle takes 2-4 years and who are therefore falling behind, and independent entrepreneurs with small teams and no compliance or legal department, who are getting the biggest efficiency gains.
- You can't just think out loud with an agent: it's super proactive and immediately starts implementing whatever you say. You need to explicitly state in the prompt whether it's a brainstorm or a concrete task.
- The cost of cognitive work has dropped sharply, which makes it possible to knock out a long-postponed backlog of nice-to-have ideas in days instead of months, especially in small businesses that software has never properly served.
- The roles most at risk are the ones built around coordination and holding shared context across a large structure (project and program managers), because autonomous teams working with agents need far less coordination between people.
- Every person is increasingly becoming not a single specialist but a small team consisting of themselves plus a set of agents uniquely configured around them, and those agents and the software built for them become part of a personal capital that travels with you into any new project or company.
The full transcript
The episode is recorded in Russian, and the full transcript of the conversation is on the Russian page of this episode. English and Spanish subtitles for the strongest episodes are rolling out on YouTube in waves.
Open the transcript