Building with Agentic RAG and Cohere

preview_player
Показать описание
First, we’ll deep-dive into agentic RAG with Cohere models, covering the transition from vanilla to agentic RAG. Then we'll show how an agent can dynamically route queries to the best data sources for smarter, faster retrievals. Using a notebook example walkthrough, participants will learn to set up and run a simple agent that powers context-aware applications.


#MicrosoftReactor #learnconnectbuild #makeazureaireal

00:00:05 - Introduction and welcome
00:02:35 - Agentic RAG and Cohere
00:02:53 - Single-shot RAG is not enough - user examples
00:05:20 - Agentic RAG is an expansion
00:08:07 - Tools in Agentic RAG workflow
00:10:45 - Question from the audience - keeping user information confidential
00:13:11 - Question from the audience - graphic/PDF support
00:14:51 - Question from the audience - R+ model example
00:17:20 - Example notebook
00:22:13 - Tool selections, instructions to agent
00:30:01 - Step-by-step breakdown
00:37:29 - Human monitoring for LLMs
00:41:00 - Question from the audience - limitation for tools?
00:45:23 - Dealing with multiple results
00:49:20 - Citations
00:56:09 - Resources, summary, what's next


[eventID:23815]
Рекомендации по теме