Agentic Development Part 2 - hands on

Value of the Information & Strength of the Argument

The video provides valuable practical insights into using AI coding agents in real-world scenarios. The host demonstrates multiple tools and workflows, sharing his personal preferences and lessons learned. The argumentation is based on direct experience, with honest acknowledgment of limitations and uncertainties. The value lies in the concrete examples and the discussion of strategies for managing multiple agents, such as using specs, todos, and PR-based reviews. The host’s approach is pragmatic and adaptable, making it useful for developers looking to integrate AI agents into their own workflows.

Scientific Rigor, Source Quality, Title Accuracy

The video is a tutorial based on the host’s personal experience, not a formal scientific presentation. It does not cite external sources, but it references specific tools and platforms (Claude Code, GitHub Copilot, etc.) which are well-known in the developer community. The title accurately reflects the content, as it is a hands-on demonstration. The rigor is moderate: the host shows real examples and discusses potential pitfalls, but the information is anecdotal and not systematically validated. The lack of formal citations is typical for this type of content, but the practical nature adds credibility.

The title accurately reflects the content: a hands-on follow-up to a previous presentation on agentic development.

7/10

The video is a practical demonstration by an experienced practitioner, showing real workflows with multiple AI coding agents. The information is based on direct experience and live examples, but lacks formal citations or rigorous scientific validation. The speaker acknowledges limitations and uncertainties, which adds credibility.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and overview of the video's purpose.
• 02:00Demonstration of multiple coding agents running in parallel.
• 05:00Explanation of using Codex in the browser and setting up a PR.
• 10:00Discussion of Cursor and its background agents feature.
• 15:00Comparison of different AI-centric IDEs and extensions.
• 20:00Explanation of the spec-driven workflow using Kira.
• 25:00Discussion of managing multiple agents and avoiding merge conflicts.
• 30:00Importance of todos for keeping agents on track.
• 35:00Overview of the PR-based review process and automated testing.
• 40:00Final thoughts and Q&A session.

• Claude Code — Mentioned as a primary coding agent used in the demonstration.
• GitHub Copilot — Used to generate code and create a pull request.
• Qwen Coder — Mentioned as a new coding agent with a large model.
• Codex — Used in the browser to create a PR.
• Cursor — AI-centric IDE fork of VS Code, used for background agents.
• Windsurf — Another AI-centric IDE fork, mentioned as an alternative.

• Claude Code — The video demonstrates Claude Code in action, consistent with its advertised capabilities.
• GitHub Copilot — The video shows Copilot generating a PR, aligning with its features.