Build Your First AI Agent with Open Source
summary
Builds a local Chilean wine sommelier agent using Olama Llama 3.1 and the Strand Agents Python library, loading a small local dataset and adding two callable tools (a wine cellar lookup and food-pairing). Uses file-based session memory so recommendations respect prior user preferences; model runs locally (zero token cost) but can hallucinate, and swapping to AWS Bedrock (Cloud Opus 4.6) improves speed and accuracy.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 2:53What is an AI agent
- 7:08Tech stack
- 10:24Workshop phases
- 13:27Project setup
- 15:38First basic agent
- 17:08Real-time responses
- 18:11Customize with system prompt
- 21:10Turn into tool-using agent
- 24:46Define tools in code
- 25:48Add pairing tool
- 28:44Session memory
- 31:34Save user preferences
- 34:38Switch to cloud model
- 37:16Cloud model demo
- 39:27Use cases and conclusions
- 41:12Challenge and giveaways
related talks
02h 08mIntroduction to AI Agents
29mDemystifying AI Agents with Python Code
24mYour First JavaScript AI Agent
33mEvaluating Your AI Agent
14mIntroducing Strands Agents, an Open Source AI Agents SDK
40mAgents in Action: Advancing Open Source AI Missions
30mMind the Gap - A Developer's Roadmap To Building AI Agents
01h 23mBuilding an AI Travel Agent That Never Hallucinates