Building an AI Agent from Scratch — RAG · Multi-Agent · MCP
Build an AI agent from scratch using Python and a single while loop, without any frameworks. In the introductory section, you’ll add tool design, failures and costs, memory, human approval, evaluation, and a web interface one by one to complete a small coding agent. In the hands-on section, you’ll implement RAG, multi-agent systems, and an MCP server yourself using real business reports from DART, then compare them numerically with LangChain and LangGraph.
58 learners are taking this course
Level Basic
Course period Unlimited
A hands-on section will be added — RAG · Multi-Agent · MCP · LangGraph
Hello, this is Dante.
A practical section will be added to this course. Building on the agent loop created in the introductory section, we will create an AI investment research workbench using actual business reports from listed companies (DART) and cover the following topics.
- Practical Tool Design: Expanding 15 disclosure lookup tools to 40, tool search, and caching
- Context engineering: handling long documents and long-running tasks, skills and sub-agents
- RAG: BM25 · hybrid · reranking, adding sources and cross-checking numbers, comparison of LangChain · LlamaIndex
- Multi-agent: Corporate research team, citation verifier, migrating to LangGraph
- MCP: Implementing a server directly → official SDK → remote deployment and token authentication → security → connecting to LangGraph
- Capstone: Completing the Workbench with FastAPI · React and the Final Grade Report
We’ll proceed by directly measuring everything numerically, and the 23 lessons will be released sequentially across 7 sections.
Those already enrolled can continue watching without any additional payment.
Thank you.




