Les OS Agentiques expliqués en 13 minutes (Sans Hype)

Les OS Agentiques expliqués en 13 minutes (Sans Hype)

Agentic OS explained in 13 minutes (No Hype)

by Erwan | IA

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Erwan | IA 👥 15K 📅 May 13, 2026 ⏱ 13 min 👁 9K 📄 expert opinion 🧭 2026-09-07
Available in: English (current) Français

Keywords

agentic OSAI agentsobservabilitymemoryorchestrationsecurityOpenClawHermes Agent

Summary

The video explains the concept of an agentic operating system (OS) as the third level of AI usage, above simple chatbots and task-specific agents. It uses the metaphor of a digital butler that continuously observes, remembers, and orchestrates specialized sub-agents to handle tasks autonomously. The creator identifies three pillars of an agentic OS: observability, memory (in three layers: short-term, long-term, and persistent personal context), and orchestrated architecture. He presents three practical paths to adopting such a system: OpenClaw (a ready-made catalog with 370k GitHub stars and 44k+ skills), Hermes Agent (which learns by observing the user for 7-20 days and replicates 70-80% of workflows), and building a custom system using Markdown files. Key technical concepts include the agentic loop, the orchestrator agent, memory layers, tools (including MCP and skills), and an event bus that enables proactivity. The video emphasizes security concerns such as prompt injection, memory exfiltration, and supply chain attacks, and recommends four guardrails: sandboxing, least privilege, auditing, and human-in-the-loop. It concludes by suggesting that while these tools are not yet native in mainstream OSes, experimenting now could provide a significant advantage in the near future.

187 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides a valuable conceptual framework for understanding agentic OS, clearly differentiating it from simpler AI tools. The argumentation is structured and logical, using the butler metaphor effectively to illustrate the continuous, proactive nature of such systems. The creator presents three adoption paths with balanced pros and cons, and emphasizes practical security measures. However, the video relies heavily on anecdotal evidence and the creator’s own experience, with limited reference to external research or benchmarks. The claim about Hermes Agent’s 70-80% workflow replication is presented without supporting data, and the overall argumentation would benefit from more rigorous evidence.

Scientific Rigor, Source Quality, Title Accuracy

The video demonstrates a good understanding of the technical landscape, correctly identifying key components like the agentic loop, memory layers, and event bus. The creator mentions specific projects (OpenClaw, Hermes Agent) and provides GitHub star counts, but does not cite any academic papers or official documentation. The description includes links to the mentioned projects (OpenClaw, Hermes Agent) and a skills marketplace, which serve as primary sources. The title accurately reflects the content, and the video stays true to its promise of a hype-free explanation. The creator’s expertise is evident, but the lack of formal citations limits the scientific rigor.

212 words

Title / Content Match

The title accurately reflects the content: a concise, hype-free explanation of agentic OS, covering definitions, examples, and practical considerations.

Quality & Reliability

7/10

The video provides a clear, structured overview of agentic operating systems, distinguishing them from chatbots and agents. It covers key concepts (observability, memory, orchestration) and security considerations, but relies primarily on the creator's expertise and anecdotal evidence rather than peer-reviewed sources. The information is generally accurate and up-to-date as of May 2026, but some claims (e.g., Hermes Agent's 70-80% workflow reproduction) lack empirical backing.

Chapters

Cited Sources

  • OpenClaw GitHub repository — Mentioned as the first path to an agentic OS, with 370k stars and 44k+ skills.
  • Hermes Agent GitHub repository — Mentioned as the second path, an agent that learns by observing the user.
  • OpenClaw Skills Marketplace — Shown as a site listing over 52,000 skills and plugins for OpenClaw.

Concurring Sources

  • OpenClaw GitHub repository — The video's description of OpenClaw's features aligns with the project's public repository.
  • Hermes Agent GitHub repository — The video's description of Hermes Agent's learning approach aligns with the project's public repository.

Contribution & Novelties

The video offers a clear, structured introduction to agentic OS, demystifying the concept and providing practical guidance for adoption. It highlights the importance of observability, memory, and orchestration, and introduces the event bus as a key enabler of proactivity. The security section is particularly valuable, outlining four concrete guardrails. The video also encourages hands-on experimentation, positioning agentic OS as a significant step forward in AI integration.

Pour aller plus loin :

118 words

Radar Profile

The radar profile shows high scores in information quantity and technical level, indicating a dense and technically rich video. The quality of information and global reliability are slightly lower, reflecting the reliance on anecdotal evidence and lack of formal citations. Overall, the video is informative and practical, but not deeply rigorous.

Reliability 7/10