2025 DataStax Agentic Workshop

2025 DataStax Agentic Workshop

• 1. ©2025 DataStax – All rights reserved. Agentic Workshop January 30th 2025

• 2. ©2025 DataStax – All rights reserved. Your hosts today! 2 David Jones-Gilardi Developer Relations Engineer Coder/GenAI nerd Melissa Herrera Developer Relations Engineer Tech/GenAI Content Junkie

• 3. ©2025 DataStax – All rights reserved. Hacking Agents - AGENDA!!! 3 ● Hack stuff ○ WTH is Langflow? ○ Build a simple agent with tools ● Agents recap ○ What are agents? ○ What are tools? ○ Agent expectations ● Agentic RAG ○ RAG recap ○ Add RAG tool to agent ○ Resources

• 4. ©2025 DataStax – All rights reserved.

• 5. ©2025 DataStax – All rights reserved. Agents recap 5

• 6. ©2025 DataStax – All rights reserved. Agents are… 6 “...a Generative AI agent can be defined as an application that attempts to achieve a goal by observing the world and acting upon it using the tools that it has at its disposal” Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google “...AI's that can perceive, reason, plan, and act” Jensen Huang Keynote at CES 2025

• 7. ©2025 DataStax – All rights reserved. Tools are… 7 “...tools bridge the gap between the agent’s internal capabilities and the external world” Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google

• 8. ©2025 DataStax – All rights reserved.

• 9. ©2025 DataStax – All rights reserved.

• 10. ©2025 DataStax – All rights reserved. Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google

• 11. ©2025 DataStax – All rights reserved. Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google

• 12. ©2025 DataStax – All rights reserved. Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google

• 13. ©2025 DataStax – All rights reserved. Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google

• 14. ©2025 DataStax – All rights reserved. Source: Wiesinger, Marlow, Vuskovic; “Agents”, Google MCP

• 15. ©2025 DataStax – All rights reserved. Level set 15 Not all models can execute function calling Tool name and description are important Execution time may be longer than you are used to You don’t always need to build an agent

• 16. ©2025 DataStax – All rights reserved. Hack Stuff 16

• 17. ©2025 DataStax – All rights reserved. Time to get building! 17 We will be using…

• 18. ©2025 DataStax – All rights reserved.

• 19. ©2025 DataStax – All rights reserved. Langflow 19 What it IS: - An open source, visual IDE for Gen AI application development including building agents and RAG (Retrieval Augmented Generation) - Built on Langchain and Python - A flexible tool that enables customization of components and integrations to external data sources and APIs What it IS NOT: - A standalone language model or AI agent itself; rather a framework for building and managing AI workflows - Limited to single AI model or vendor; it supports multiple models and can be integrated with various AI technologies

• 20. ©2025 DataStax – All rights reserved. Getting Started 20

• 21. ©2025 DataStax – All rights reserved. Getting Started 21

• 22. ©2025 DataStax – All rights reserved. Agentic RAG 22

• 23. ©2025 DataStax – All rights reserved. RAG Recap 23 RAG = Retrieval Augmented Generation Used to provide LLMs with domain specific knowledge Can leverage vector databases or other sources Vector databases give rich results from user queries

• 24. ©2025 DataStax – All rights reserved. Resources 24 The Principles for Reliable, Safe and Fast AI Agents • rwilinski.ai Berkeley Function-Calling Leaderboard Agents (white paper by Google) How Agentic Hybrid Search Creates Smarter RAG Apps

• 25. ©2025 DataStax – All rights reserved. Get Langflow (we’d love your github stars) 25

• 26. ©2025 DataStax – All rights reserved. Make your CodeBeast! 26

• 27. ©2025 DataStax – All rights reserved. 27