Building Agentic AI applications with Positron and AWS Strands Agents
summary
Shows a working pipeline that turns Python functions and third‑party APIs into callable tools for an LLM agent using Strands Agent SDK and Strands Agent Tools, with models hosted via Amazon Bedrock (Anthropic Claude Sonnect 4 shown) or local Ollama/OpenAI. Demo wraps the agent into a Streamlit .py and runs in Positron, integrates Pinecone vector DB and DuckDuckGo web search, and uses a calculator and time/weather tools; the agent autonomously plans tool calls, answers a recent 2025 event query and computes a huge square root with correct results.
AI-written summary · about
chapters (10)
- 0:00Agentic AI with Positron and Strands
- 2:07AWS Strands Agents SDK overview
- 5:59Minimal code agents in Positron
- 7:45AI model setup for agents
- 9:12Built-in tools for agents
- 9:56Python functions as custom tools
- 12:07Third-party API integration
- 13:55Knowledge bases for enterprise data
- 14:59Deploy agents as Streamlit apps
- 15:46Summary
related talks
19mBuilding Agents at Cloud Scale
29mBuilding Production-Ready AI Agents
14mIntroducing Strands Agents, an Open Source AI Agents SDK
21mServerless Agents: Real-World Tooling with Strands SDK, MCP & AWS
28mBuilding Agentic AI applications on AWS
30mBuilding Robust, Secure LLM and Agentic AI Workflows
40mBuilding Intelligent Search Systems With Agentic RAG and OpenSearch
40mAI Agents in Practice