Did we forget Step Functions over Durable Functions?🚀☁️ #99

Step Functions adds in 28 new service integrations (AgentCore included)

Mar 30, 2026

In the previous issue, OpenClaw on Lightsail🚀☁️ #98 focused on the Lightsail supporting OpenClaw as a managed offering, making it even simpler.

Step Functions just became a control plant for AI Agents with the new integrations of AWS Services and API Actions, which include Bedrock AgentCore and also Durable Functions, bringing orchestration to AI Agents that can be provisioned as part of the state execution and also in conjunction with Map State, ensuring the Agentic workflows are executed in parallel.

The integration also has S3 Vectors enabling your RAG pipelines on autopilot with the ingestion of data into knowledge based with the “middle Lambda fn”.

While we were busy with Durable Functions, Step Functions made it possible to trigger a durable function, bringing the “super manager” orchestrating the durable functions (the specific workflow).

Is AI now predicting the future? Johannes Koch (AWS DevTools Hero)’s Nexus Share did predict that the Serverless Terminal issue 99 would be on Step Functions!

Nexus Share AI is making predictions that are coming true!

⭐ Pick of the month

This month’s Pick is “Amazon S3“.

Anmol Shukla dives into designing a dynamic serverless data pipeline on AWS which leverages S3 for data storage, AWS Glue Crawlers for schema discover and the whole EDA and CDK infrastructure.

Dania Alrefai gets into the architecture designing RAG systems on AWS with S3 Vectors and how S3 Vectors plays right over traditional approach along with the best of hybrid - S3 Vectors and a vector database.

Jones Zachariah Noel N (author)’s blog Amazon S3 is more than storage and brings in a lot for the analytics ecosystem reflects on how S3 has now grown to be a core compnent for data with S3 Tables.

🚀 AWSome content to learn from

Darryl Ruggles writes about Powertools for AWS Lambda - Best Practices By Default bringing the structure to your logs, metrics and stiching the trace gaps to also a better way to handle idempotency.

Vadym Kazulkin as part of the multi-part series on Serverless applications on AWS with Lambda using Java 25, API Gateway and DynamoDB from the basic app to also performance benchmark along with SnapStart.

Elena van Engelen-Maslova explains how you can mock the Serverless app end to end with goodbye flaky external APIs, Hello mocking in the cloud with MockNest Serverless bringing WireMock-compatible API mocking to Lambda.

Chibuike Nwachukwu shares how to build a sercure serverless document management system that supports RBAC and ABAC using Amazon Cognito and Amazon Verified Permissions via Cedar Polices.

Darryl Ruggles neatly explains how AWS Lambda now knows where it lives: exploring the new AZ Metadata with the routing and cutting AZ data transfer costs along with how it brings in resilience for Lambda.

Aditya Ganti demysfies what really happens under the hood of a Lambda function with the Firecracker instance and how it all boots up during the Lambda init phase.

Kenton Varda, Sunil Pai, and Ketan Gupta explain the internals of Sandboxing AI Agents, 100x faster with Cloudflare Dynamic Worker Loader along with a walkthrough of the code and tips with Cloudflare MCP server.

Daniel Grzelak’s “But without PassRole it should be fine”, Lambda edition exposes the basic lambda:UpdateFunctionConfiguration IAM PassRole and how it can turn a disaster for you.

Ken Winner shares about using the CDK Toolkit diff to share diffs, how to bring destructive changes to the surface before they hit production and also having it as a report in your PRs.