# Intro to Agents Syllabus In this course you will: - Use the Basic Agency Pattern with the OpenAI Agents SDK (& discuss other frameworks) - Write an effective system prompt for complex problem solving or process rubrics - Write a semantic chunking summarizer for extremely long documents - Write Agents who can retrieve knowledge from documents - Write Agents who can reflect on their actions and take notes - Create a Swarm of agents that can coordinate on a goal <p><a href="https://www.loom.com/share/661b73af8f7541518eb7e1dfa4dc0437">▶︎ Intro & What Agents Can Do 8.3.26</a></p> **Full-day recordings** — newest first - **Tue 15 Sep 2026** *(most recent)* · [▶︎ Recording](https://drive.google.com/file/d/16OYwGcIN3qSdqC1Bk01bB6ZqsYU6sIxu/view) - **Mon 3 Aug 2026** · [▶︎ Recording](https://drive.google.com/file/d/1Y48IS9RGq0JtUPi3RhNBw6ePbgYTFVqG/view) - **Mon 13 Apr 2026** · [▶︎ Recording](https://drive.google.com/file/d/1Tm8WYlJ1dsjjm7rV25JxvmNFu-Bwbp6N/view) - **Mon 23 Mar 2026** · [▶︎ Recording](https://drive.google.com/file/d/1wimZr0MSpJG6g1NVD4d6JnhXuDflKxEs/view) - **Fri 30 Jan 2026** · [▶︎ Recording](https://drive.google.com/file/d/1f_uYdZrzCMD_F0lWR9bXH6NUVCaMviqn/view) ## Course Pages - [Request-Response Agents](/za8X9-feQ8mfgRwXSdPC7w) — system prompts, the basic agent pattern, ReACT, OpenAI Agents SDK - [Request-Response Agents - Exercises](/eCaj346SQnWLkvmhnISpxA) - [The Basic Agent Pattern](/CrOL86lbT86nlgbDNr0yoA) - [ReACT Agency Pattern](/lgxLVO1WQOOBZ1lB5zp25Q) - [ReAct with Python Prompts](/tjv3zaffTdKvxJMTCv95sg) - [Retrieval-Augmented Agents](/_2c0gcJmTlGQoZKFqG1JnQ) — prepare data, load into Postgres Vector DB or FAISS, build a RAG pipeline - [Resources](/AzYfjBtCSteXhQZVM7XbrA) - [Retrieval-Augmented Agents - Resources](/QskFVvxhRJqj-KKaQLnEFg) - [Retrieval-Augmented Agents - Task Checklist](/rX8VFig4RGWsq_IfS-DEZA) <!-- rag-fundamentals:start --> - [RAG Fundamentals — full sequence](/KdEjARFXQCymcBYSkua40Q) — *from Context Engineering:* what RAG is, the pipeline, four chunking strategies compared, two exercises and a checklist - [What is RAG?](/UxSSdZ5-SDyyTjABg-O0Aw) - [Embeddings — the Pimantle connection](/9EeXgrePQCqUBFwmiwUbCg) - [The RAG pipeline](/EmikPqMzSryT-5-o_7XToA) - [Why chunking matters](/bhKmS93_Rx-OFdWBO1HBjQ) - [Choosing a chunking strategy](/Sf-OCmL3RsyyOkY2rOAI0w) - [Exercise — build a basic RAG system](/25tkGWMMSZ2PrVp9APOmMw) - [Exercise — chunking comparator](/tACHUH1LQvmy9ywSevm1kQ) - [Key takeaways](/iCIfOZe2QHOmvaJ4qtlV1Q) - [Advanced RAG — the deeper course](/r9jn1iULQpKp2gT03GLumg) — vector database internals, performance, neural rerankers - [Advanced RAG — slides](https://docs.themultiverse.school/p/UTC3dcQPQpeDN_GSWysVLQ) <!-- rag-fundamentals:end --> - [Building Your Own Agent Framework](/zqOv-NhzR9227LazFvXAOw) — make our own agents framework in ~140 lines of code - [Exercises](/UQLolQAIQp2PTW9y0mmWMw) - [Building Your Own Agent Framework - Resources](/Uu4i3J0jT-6PfkS3zKoZ_w) - [Key Takeaways - Building Your Own Agent Framework](/QsirnPeyTGm0E0hE93sjpA) - [Autonomous Agents Workshop 1](/gtV4bVU-SbaC6MCGavJRgg) — goals instead of tasks, tool use, async operation - [Autonomous Agents Workshop 1 - Exercises](/1qmjWI03TeSDbrC58kS4MA) - [Autonomous Agents Workshop 1 - Resources](/lYPS6bbBTAqHa9XViHHuTQ) - [Autonomous Agents Workshop 2](/Chzn5QJkT_iAP2tNASe6eQ) — managing a team of long-horizon agents - [Discussion Questions](/Z9jM8FUZSpiXmqk_f1hIyQ) - [Multi-Process Agents](/FCBlk9_ERJK9gkBxy2ni3w) — orchestrating a multi-process agent swarm - [Multi-Process Agents - Framework Comparison](/nKmLsTznRM2i2rYVwyCBhA) - [Multi-Process Agents - Resources](/lgJN78k4QuS9gWSflxDSfA) - [Swarm Prototyping](/I199hItQQlypL7coe8mclg) — multiple agents acting as a single agent - [Swarm Prototyping - Resources](/3JvXpzx0S8K_gV7zU0Rkkg) - Swarm Recruiting *(coming soon)* — reuse existing agents (or tweak their prompts) for new tasks - [Final Agents Wrap-up](/R1Rea8quSFmiYrpnofo2LA) - [Security and Control Layers](https://docs.themultiverse.school/s/PU_-fH-RP) — Intercept MCP, Langfuse tracing, agentic AI security - [Production Agent Engineering — the deeper course](/5hJRC0luR8C-QMcSlvUpiw) — orchestration patterns, self-critique loops, durable memory, open-source backends ## More recordings for this class - [Hire your first agent](https://www.loom.com/share/1a5290d497ca4d20a9b9690c23ea1b44) — Define an agent's role, goals, and success metrics as a concrete job description - [Understanding Agents and Environments](https://www.loom.com/share/d7fda183f7bd4dd7a3d80560b5ab39c8) — Model an agent as an OODA loop and test it inside a prebuilt Gymnasium environment - [Building the Agentic Loop](https://www.loom.com/share/079bb7c17d794d73ab2c9bdba0bf9252) — Code a stateful conversational loop that switches between cloud and local AI models - [Agent Planner Tutorial](https://www.loom.com/share/1057fe0f54f747ea882ffe9c180679e5) — Orchestrate a multi-agent swarm using the OpenAI Assistants API and custom functions - [Exploring OpenAI's Agents API](https://www.loom.com/share/109e123d77c64073affbed3101e6ae8a) — Add custom Python functions as tools and handle the full tool-call lifecycle - [Building and Implementing AI Agents with OpenAI SDK](https://www.loom.com/share/1a2ddcc30d5c4d8dac8e3780165be0c2) — Build an agent with the OpenAI SDK, adding web search and custom file-writing tools - [Building Tools with Python 🐍](https://www.loom.com/share/1ff19718275f448281c6a91227fb6575) — Register Python functions as tools for AutoGen agents to call - [Creating an Agent Writing Agent](https://www.loom.com/share/354d728804be46b7ae2fe8270a4e427f) — Build two OpenAI Assistants that plan and write system prompts for an agent swarm