Agent Speedrun: AWS workshop builds and ships a customer service agent with Strands Agents
AI Engineer · youtube · 2026-10-11
At the AI Engineer World's Fair, AWS's Elizabeth Fuentes Leone and Sandhya Subramani run a hands-on workshop building a production-bound customer service agent that looks up accounts, checks orders, and handles refunds — with the hard part being control between steps and keeping it in production.
Walkthrough
- Starting from a system prompt and three tools backed by mock customer data using Strands Agents, inspecting messages and tool calls
- Adding a hook-based tool invocation rate limiter for programmatic control, and clarifying the precise semantics of blocking a tool call vs. ending the agent's turn
- Skills and steering: markdown instructions deliver troubleshooting, order tracking, and refund procedures on demand, while steering handlers check workflow order and response tone, contrasting programmatic checks with instructions left to a probabilistic model
- Deploying the same code to Amazon Bedrock AgentCore: entry point, local testing, runtime setup, generated infrastructure, and observability for sessions, traces, and token use
Full timestamps cover the path from notebook agent to deployed service; Strands Agents and AgentCore sample repos are linked.
More from coding & agent
- Jose Valim shows Campfire AI benchmark was rigged: Elixir faced far stricter checks than Go/Rust — zeeg · 2026-10-12
- Ask the model for bullet points, write the changes yourself: keeping your voice with AI — ctjlewis · 2026-10-12
- Outer loop vs inner loop: why your context-fetching agent shouldn't do dev, review, or verification — alexcovo_eth · 2026-10-12
- Streamer runs 25 Opus agents on live chat, ships 90 overnight changes — TheMoonMidas · 2026-10-12
- 15 Terms Every AI Product Builder Should Know, From Harness to Hooks — aakashgupta · 2026-10-12
- Graph Engineering and Harness Engineering: a blueprint for building reliable agentic systems — Pavan_Belagatti · 2026-10-12