From RAG to AI Agents: Function Calling and Tool Use - Alexey Grigorev

In this 3rd workshop of our series, Alexey Grigorev, Founder of DataTalks.Club, shares his expertise in building advanced LLM applications transitioning from basic Retrieval-Augmented Generation to sophisticated Agentic RAG. We explore the evolution of AI agents, focusing on how to move beyond rigid, hard-coded pipelines to dynamic, reasoning-based workflows. You'll learn: How to turn simple search-and-answer systems into smart agents that can think for themselves. Ways to let an AI decide exactly when it needs to look up new information. How to build a "loop" that lets an AI keep working until it finds the right answer. A simple trick to help the AI remember what was said earlier in a conversation. How to automatically tell the AI what tools it has available using your existing code. How to track exactly how much money you are spending on every AI request. How to set "emergency stops" to keep your AI from running forever or making mistakes. Links: Course: https://github.com/DataTalksClub/llm-... Module 1 Part 2: https://github.com/DataTalksClub/llm-... Guardrails: https://aishippinglabs.com/workshops/... TIMECODES: 00:00 Agentic RAG and LLM Zoomcamp 03:15 Environment setup with GitHub Codespaces and UV 06:45 Analyzing the FAQ chatbot and original RAG workflow 13:05 Transitioning from rigid flows to agentic reasoning 17:40 Implementing function calling for dynamic search 21:20 Defining language agnostic tools using JSON schemas 25:10 Processing tool calls and local search execution 28:45 Managing statelessness with conversation history and memory 32:30 Monitoring token usage and API cost calculation 36:00 Building the agentic loop for non-deterministic flows 44:10 Refining loop logic for multiple tool iterations 52:15 Evaluating agent frameworks and the Toy AI Kit 56:45 Inferring tool schemas from docstrings and type hints 1:01:25 Implementing safety measures and loop exit conditions 1:14:15 Choosing between agents and standard RAG pipelines This workshop is designed for AI Engineers, Data Scientists, and Software Developers who want to master the architectural shift from simple RAG to agentic systems. It is ideal for practitioners looking for framework-agnostic principles to build scalable and cost-effective LLM applications. Connect with DataTalks.Club: Join the community - https://datatalks.club/slack.html Subscribe to our Google calendar to have all our events in your calendar - https://calendar.google.com/calendar/... Check other upcoming events - https://lu.ma/dtc-events GitHub: https://github.com/DataTalksClub LinkedIn -   / datatalks-club   Twitter -   / datatalksclub   Website - https://datatalks.club/ Connect with Alexey Twitter -   / al_grigor   Linkedin -   / agrigorev   Check our free online courses: ML Engineering course - http://mlzoomcamp.com Data Engineering course - https://github.com/DataTalksClub/data... MLOps course - https://github.com/DataTalksClub/mlop... LLM course - https://github.com/DataTalksClub/llm-... Open-source LLM course: https://github.com/DataTalksClub/open... AI Dev Tools course: https://github.com/DataTalksClub/ai-d... 👉🏼 Read about all our courses in one place - https://datatalks.club/blog/guide-to-... 👋🏼 Support/inquiries If you want to support our community, use this link - https://github.com/sponsors/alexeygri... If you’re a company, reach us at [email protected] #agenticai #rag #aiagents #llm #dataengineering #python #aiengineering #llmzoomcamp #functioncalling #aiarchitecture #opensourceai #machinelearning #generativeai #naturallanguageprocessing #semanticsearch #vectorsearch #aiworkflow #aitechnology #aisoftware #datatalksclub