AWS
+1
Jun 16, 2026
•
8 min read
SL#73 - AWS AI Series (13/30) - Multi-Agent Patterns in Strands: Agents-as-Tools, Swarm, and Graph (and When Each Is Wrong)
Build the same research/writer/reviewer pipeline three ways with the Strands Agents SDK, then learn which pattern to actually reach for.
AWS
+1
Jun 15, 2026
•
9 min read
SL#71 - AWS AI Series (12/30) - Agents You Can Audit: Lock an AgentCore Agent Behind JWT Auth and Trace Every Call in CloudWatch
We take a Strands agent on Bedrock AgentCore Runtime, put a Cognito JWT wall in front of it so only authenticated callers get in, and wire OpenTelemetry so every invocation shows up as a traced session in the CloudWatch GenAI Observability dashboard.
AWS
+1
Jun 14, 2026
•
8 min read
SL#70 - AWS AI Series (11/30) - Give Your Agent Hands: Build a Fact-Checker With AgentCore Code Interpreter and Browser
Build an agent that browses a live page for a real number, then writes and runs Python in a sandbox to verify a claim about it. Two AgentCore built-in tools, one Strands agent.
AWS
+1
Jun 13, 2026
•
9 min read
SL#69 - AWS AI Series (10/30) - Turn a Lambda Into MCP Tools Your Agent Can Call With AgentCore Gateway
Build a managed MCP server in front of an AWS Lambda function, expose two tools to a Strands agent without writing protocol code, then add a REST API as a second target.
AWS
+1
Jun 12, 2026
•
9 min read
SL#67 - AWS AI Series (9/30) - Agents That Remember: Persist User Preferences Across Sessions with AgentCore Memory
Build a Strands agent that recalls who the user is and what they like across separate runs, using Amazon Bedrock AgentCore Memory for both the conversation buffer and the long-term preference store. Plan for about 90 minutes if you finished episode 7, two hours if this is your first Strands agent.
AWS
+1
Jun 11, 2026
•
11 min read
SL#66 - AWS AI Series (8/30) - From Laptop to Prod: Deploy Your RepoScout Agent to Bedrock AgentCore Runtime
Take the local RepoScout agent from episode 7 and ship it to a managed, isolated microVM runtime with the AgentCore harness and CLI - the agent code barely changes.
AWS
+1
Jun 10, 2026
•
10 min read
SL#65 - AWS AI Series (7/30) - Strands Agents 101: Build a Code-Defined Agent With 3 Custom Tools in ~50 Lines of Python
We build a local agent that answers questions about any codebase on your disk, using three custom tools and Amazon Bedrock. Plan for about 60 minutes if you have AWS credentials ready, 90 if you have to enable Bedrock model access first.
AWS
+1
Jun 8, 2026
•
9 min read
SL#62 - AWS AI Series (5/30) - Make a Small Model Act Big: Fine-Tune Nova Micro on Bedrock and Benchmark It vs Base
Take the smallest Nova model, teach it one job, and watch it jump from 41% to 97% accuracy for $2.18 - then serve it at the cheap-model price.
AWS
+1
Jun 7, 2026
•
9 min read
SL#61 - AWS AI Series (4/30) - Stop Eyeballing Prompts: Version a Prompt and Run a 3-Model Eval Job in Bedrock
We'll move one prompt out of your codebase into Bedrock Prompt Management, version it, then run an LLM-as-a-judge evaluation job that scores three models against your own dataset and tells you which one to ship. About 90 minutes, most of it waiting on the eval job.
AWS
+2
Jun 6, 2026
•
8 min read
SL#59 - AWS AI Series (3/30) - Bedrock Guardrails: Build a Guarded Chatbot Endpoint That Actually Blocks the Attacks
Build a support-assistant endpoint protected by Amazon Bedrock Guardrails, then run four attacks against it and watch each one get blocked or masked. Plan for 60 to 90 minutes if you have used boto3 before.
AWS
+1
Jun 5, 2026
•
9 min read
SL#58 - AWS AI Series (2/30) - Production RAG in an Afternoon: Build a Bedrock Knowledge Base on S3 Vectors
Build a queryable, citation-backed knowledge base over your own documents with Bedrock Knowledge Bases and S3 Vectors, entirely from Python. 60-90 minutes, under a dollar in AWS charges.
AWS
+2
Jun 4, 2026
•
8 min read
SL#57 - AWS AI Series (1/30) - One API, Two Model Families: Build a Bedrock CLI with Converse, Streaming, and Validated JSON
Episode 1 of the AWS AI Series: a Python CLI that talks to Amazon Nova 2 and Claude Sonnet 4.6 through one identical code path, streams tokens, and returns schema-validated JSON. About an hour, end to end.
