Agent Speedrun — Elizabeth Fuentes Leone & Sandhya Subramani, AWS
Source
youtube.com
Author
AI Engineer
Date
Why it matters
Contrasts programmatic hooks and steering with prompt-only instructions and shows the path to a deployed agent with tracing. Helps engineers decide where to enforce control in an agentic loopThe cycle an agent runs in: decide, call a tool, read the result, decide again — repeating until the goal is met or a stop condition fires.Full definition →.
Key takeaways · AI-distilled
The build starts with a Strands Agents system promptThe standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.Full definition → and three tools (account lookup, order check, refund) over mock customer data, then adds a agent hookA script the harness runs automatically at a fixed point in an agent's loop — before a tool call, after an edit, on session start.Full definition → that caps tool invocations; lowering the cap makes the limit visible in a live test.
An audience question exposed an ambiguity: a limit can block one more tool call or end the AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition →'s turn entirely. The presenters use it to show that a control needs a precise definition of what it stops.
Skills are markdown procedures for account troubleshooting, order tracking and refunds that load only when needed, while separate steering handlers check that the workflow runs in order and that responses keep the right tone.
Deploying to Amazon Bedrock AgentCore reuses the same agent code with a runtime entry point, local testing, generated infrastructure and an observability configuration, which is what makes sessions, traces and token use inspectable.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
agent hook — A script the harness runs automatically at a fixed point in an agent's loop — before a tool call, after an edit, on session start.
agentic loop — The cycle an agent runs in: decide, call a tool, read the result, decide again — repeating until the goal is met or a stop condition fires.
system prompt — The standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.
videoAI Security Engineer Foundations + Certificate — Micah Silverman, Snyk
videoThe SDLC You Build Yourself — Andy Wong, Andrei Bocan & Shane Wolf, Atlassian
videoCan Your Agent Hear You Now? Building Live Voice Agents with Gemini — Thor Schaeff
videoBuilding Self-Learning Loops for Your Agent — Fuad Ali, Arize AI
