Developers can now stitch together a full‑stack AI assistant on Amazon’s cloud using the open‑source Strands framework and the managed Agent Core service. The tutorial walks through a Python‑based educational bot that selects a Bedrock model, receives a system prompt, and returns answers to queries in physics, chemistry, math or geography. Minimal code—just a few boto3 calls—spins up a functional agent.

Agent Core supplies the runtime, memory, and observability layers that let the Strands agent run reliably at scale. Its Runtime isolates sessions, streams responses, and supports HTTP or MCP protocols, while the optional Memory store extracts durable facts across conversations. Additional modules—Gateway, Identity, Policy, Browser, Code Interpreter—can be attached as the bot’s responsibilities grow.

To follow the guide, users install the AWS CLI, CDK, and the Agent Core CLI, then configure credentials and select a Bedrock model such as Claude Sonnet. By separating agent logic (Strands) from deployment concerns (Agent Core), the pattern mirrors serverless functions but adds conversational state and tool access. The result is a production‑ready, cloud‑hosted AI assistant that can be extended with memory or external APIs.