LLM Agents Podcast Summaries

LLM Agents on Yedapo: 3 summarized podcast and YouTube episodes. Each includes key takeaways, core concepts and notable quotes with timestamps.

I finally found a reason to build my own AI agent
Dreams of Code
May 19, 2026
By treating broken builds as low-value tasks rather than technical hurdles, the host built a custom agent system to autonomously fix cross-platform regressions. Running on local hardware and secured by Tailscale, these agents diagnose failures and submit pull requests, effectively eliminating the context-switching tax associated with managing Windows and Linux builds from macOS.
Key insight: Rather than relying solely on generic coding harnesses, the host used the Rust 'Cersei' crate to build a hybrid agent that uses deterministic code for standard API tasks and LLMs only for diagnosing and patching logic, ensuring higher reliability and better observability.
The Biggest Problem Of AI Coding Is Finally Solved
AI LABS
May 8, 2026
As AI-generated code accelerates development, security risks have surged. Vercel's DeepSeek provides a structured, parallelized security harness that uses LLM agents to systematically scan and validate large repositories, effectively reducing false positives compared to standard ad-hoc agent reviews.
Key insight: DeepSeek uses a two-stage architecture—a preliminary regex filter followed by parallelized agent analysis—to optimize token usage and prevent security review timeouts in massive codebases.
Burn through the backlog from hell with /triage
Matt Pocock
May 7, 2026
Effective AI agent workflows require structured task management to prevent inefficiency. By implementing a state-machine based 'Triage' skill, developers can transform chaotic GitHub issues into actionable, high-quality briefs that AI agents can execute autonomously.
Key insight: Managing AI agents is primarily a queue management problem; the 'ready for agent' label acts as a vital gatekeeping mechanism to prevent AI from attempting low-quality or out-of-scope tasks.