Comment j'ai créé un assistant qui me connait par coeur

Comment j'ai créé un assistant qui me connait par coeur

by Eliott Meunier

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Eliott Meunier 👥 51K 📅 February 17, 2026 ⏱ 25 min 👁 10K 📄 tutorial 🧭 2026-08-27
Available in: English (current) Français

Keywords

AI memorycontextObsidianClaude Codeterminal

Summary

Eliott Meunier presents a system for creating a personal AI assistant that retains full context across sessions. He criticizes the built-in memory of ChatGPT and Claude as noisy and unstructured, advocating instead for a structured approach where the AI operates within a well-organized local knowledge base. The system comprises three layers: storage (Obsidian with Markdown files), the AI agent (Claude Code), and the interface (terminal). He explains why Obsidian is superior to Notion or Evernote due to its local, universal Markdown format and extensibility. Claude Code is highlighted as a powerful agent capable of reading, writing, and executing commands on the computer, far surpassing web-based chat interfaces. He demystifies the terminal with four essential commands (pwd, ls, cd, ..) and demonstrates live how the AI can manage tasks and even create a website. The video concludes with installation steps and a call to join a webinar for a full system walkthrough.

151 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video offers valuable practical insights into building a context-aware AI assistant, emphasizing a systemic approach over isolated tools. The argumentation is coherent and persuasive, supported by live demonstrations that illustrate the capabilities of the proposed system. However, the claims about AI memory limitations and the superiority of Claude Code are based on personal experience rather than empirical data, which weakens the scientific rigor. The tutorial is well-structured, progressing from problem to solution, and provides actionable steps for implementation.

Scientific Rigor, Source Quality, Title Accuracy

The video lacks formal scientific sources, relying instead on the creator’s expertise and product documentation. The description provides links to a webinar and bootcamp, which are promotional rather than academic. The title accurately reflects the content, and the video is well-organized with clear chapters. The demonstrations are credible but not independently verifiable. Overall, the rigor is moderate, typical for a practical tech tutorial.

157 words

Title / Content Match

The title accurately reflects the content: the creator demonstrates how to build a personal AI assistant with persistent context.

Quality & Reliability

7/10

The video provides a practical, hands-on tutorial with live demonstrations, but relies heavily on personal experience and opinions rather than verifiable scientific sources. The claims about AI memory limitations and the superiority of Claude Code are plausible but not backed by empirical evidence.

Chapters

Cited Sources

Concurring Sources

Contribution & Novelties

The video contributes a practical methodology for integrating AI into a personal knowledge management system, emphasizing the importance of structured context over built-in memory features. It offers a clear three-layer architecture (storage, AI, interface) and demystifies the terminal for non-technical users.

Pour aller plus loin :

79 words

Radar Profile

The radar profile shows high scores in information quantity and quality, reflecting the tutorial's rich content and practical value. The technical level is moderate, accessible to beginners, while reliability is slightly lower due to the lack of formal citations. Overall, the video is a solid practical guide.

Reliability 6/10

💬 No comments were provided for analysis.