Claude a changé, faut se mettre à jour

Claude a changé, faut se mettre à jour

Claude has changed, time to get up to speed

🎙 Renaud Dékode 👥 249K 📅 August 29, 2026 ⏱ 48 min 👁 38K 📄 tutorial 🧭 2026-09-07
Available in: English (current) Français

Keywords

ClaudeAnthropicmemorybrowsermodels

Summary

In this video, Renaud Dékode provides a comprehensive update on the major changes to Claude, the AI assistant by Anthropic, during the summer of 2026. He emphasizes that many users are still using old habits, leading to frustration with the new models. The video covers five key changes: the new memory system, Cowork, the integrated browser, Gmail autonomy, and the new role distribution among Opus 5, Sonnet 5, and Fable 5. He explains that the memory feature is more than just storing facts; it learns from conversations and can influence Claude’s behavior, including sensitive topics. The integrated browser allows Claude to interact with web pages directly within the app, offering speed and convenience, but it lacks the security of using the user’s own browser. He also discusses the new autonomous capabilities with Gmail, where Claude can send emails on its own, but warns about the need for control. Finally, he clarifies the new model hierarchy: Sonnet 5 is now the go-to for daily advanced tasks, while Opus 5 is more specialized and autonomous, requiring a different prompting approach. The video aims to help users adapt to these changes and use Claude effectively in 2026.

194 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable, hands-on insights into Claude’s recent updates, based on the author’s direct testing. The argumentation is clear and structured, with practical examples. The author emphasizes the importance of understanding the new memory system and the shift in model roles, which are crucial for effective use. He also addresses common user frustrations, attributing them to a misunderstanding of the new features rather than a decline in quality. The reasoning is logical and supported by demonstrations, though it relies heavily on personal experience rather than external data.

Scientific Rigor, Source Quality, Title Accuracy

The video is a tutorial based on the author’s personal experience, with no external sources cited. The information is presented transparently, noting that some features are still rolling out and may not work perfectly. The title accurately reflects the content, and the video stays on topic. The author’s credibility is established through his channel’s focus on AI, but the lack of citations reduces the scientific rigor. The public comments are not provided, so no analysis of audience reception is possible.

183 words

Title / Content Match

The title accurately reflects the content: a comprehensive update on Claude's recent changes, including new models, memory, and browser features.

Quality & Reliability

7/10

The video is a practical tutorial based on the author's direct experience with Claude's summer 2026 updates. It clearly distinguishes between features he has tested and those still rolling out. The information is presented with transparency about limitations and bugs, but lacks external verification or citations.

Key Moments

Cited Sources

  • Renaud Dékode's website — Author's official site with additional resources and community.
  • Klub Renaud Dékode — Paid membership for deeper AI training and community.

Concurring Sources

Contribution & Novelties

The video provides a timely and practical update on Claude’s 2026 changes, focusing on the new memory system and model hierarchy. It offers actionable advice for users to adapt their workflows. The author’s hands-on testing adds credibility, though the lack of external references limits the depth of verification.

Pour aller plus loin :

89 words

Radar Profile

The radar profile shows high scores in information quantity and quality, with a moderate technical level. The reliability is solid but not perfect, reflecting the tutorial's practical nature and lack of external citations.

Reliability 7/10