Descriptions:
Nick Saraev delivers a two-hour practitioner course on AI agents, covering both foundational concepts and advanced orchestration techniques he uses in a business generating over $4 million annually. The course is platform-agnostic — covering OpenAI Codex, Anthropic Claude Code, and Anti-Gravity — and requires no prior programming experience.
The course opens with a live demo of five Claude Code agents operating independent Chrome browsers to dynamically fill out contact forms from a leads list, illustrating the parallelization advantage of multi-agent systems. From there, Saraev walks through the core agent workflow loop before introducing techniques including: self-modifying agent instructions where agents rewrite their own rules mid-task to reduce error rates; multi-agent MCP orchestration that registers Codex, Gemini, and Claude as MCP servers within a single conversation thread; video-to-action pipelines that extract step-by-step instructions from YouTube tutorials and execute them locally; stochastic multi-agent consensus using statistical spread across parallel agent runs; agent chat rooms for structured debate to improve output quality; and subagent verification loops for real-time peer review between agents.
Saraev’s personal stack — Anti-Gravity wrapping Claude Code with Anthropic models — is demonstrated throughout. The course closes with context management strategies, output quality optimization, and token pricing, making it one of the more complete end-to-end introductions to production agentic AI available in a free format.
📺 Source: Nick Saraev · Published March 08, 2026
🏷️ Format: Course Lesson
