​Build A Full AI Agent From Scratch— RAG + MCP + Chat

​Why this is important right now

​Agents are moving from hype to workhorse. But without good fetching, correct routing and context management they become either slow, expensive, or simply wrong. This workshop teaches you exactly what separates a demo from a reliable product:

• ​RAG that actually finds things (embeddings + FAISS)

• ​MCP routing driven by semantics (dynamic tool selection)

• ​Chat memory that helps (multi-turn without losing the thread)

​Come to build something that delivers value the same evening—and that you can connect to your own datasets and APIs afterwards.

​What you learn — and why it matters

​Embeddings in practice: Choose a model, understand similarity metrics (cosine vs L2) and build a FAISS setup that scales. Why: If you don’t hit the right document quickly, the agent dies in the real world.

​MCP-driven agent selection: Represent areas of responsibility semantically and route requests where they belong. Why: Less manual wiring, fewer mis-routings, easier to extend.

​Full RAG loop with chat: Maintain context, retrieve the right knowledge, synthesize answers – and handle “uncertain” with fallback. Why: This is the core of robust agent flows in support, internal tools, and knowledge applications.

​Agenda (hands-on)

• ​Basic Vector Search (20 min)
  Embeddings → FAISS → top-k. You build ask_rag(query, k) and test.
  

• ​MCPs & Agent Selection (40 min)
  Define 3 MCPs (ticketing, user_management, notifications). select_mcp(query) chooses the right agent via cosine similarity.
  

• ​Food and drinks, on the house! (30 min)
  

• ​Full RAG + Chat (50 min)
  Chat history, dynamic MCP selection, document retrieval in the MCP, aggregate top-k and respond. Bonus: fallback agent, multi-agent collaboration, LLM summarization.
  

​For whom?

​Developers, data scientists and technical PMs who want to build useful agents—not just POCs.

There is a limited capacity for this event so apply fast to maximize your chances of securing a spot.

​You need

​Laptop with Python 3.10+

​An editor you like (VS Code is great)

​Time required

​At least 3 hours: short intro + lots of building. The pace is high, the payoff higher.

​Extra challenges (for the hungry)

• ​Dynamic chunking of long texts

• ​Multi-agent orchestration (retrieve → summarize → enrich)

• ​Visualize the flow: Query → MCP → FAISS → Response

• ​Benchmark MCP routing and top-k strategies

​More information about location and more comming soon!

​Sign up. Bring your pc. We’re building agents that actually deliver.