AWS Compute Blog

Serverless ICYMI Q3 2025

Welcome to the 30th edition of the AWS Serverless ICYMI (in case you missed it) quarterly recap. At the end of a quarter, we share the most recent product launches, feature enhancements, blog posts, videos, live streams, and other interesting things that you might have missed!

In case you missed our last ICYMI, check out previous ICYMI posts.

Figure 1: Serverless calendar Q3 2025

Figure 1: Serverless calendar Q3 2025

GOTO Serverless Bengaluru

The Asia Serverless and GenAI Tour comprised 24 events across five countries. Cities included New Delhi, Bengaluru, Singapore, Manila, Bangkok, Perth, Melbourne, and Sydney. The GOTO Serverless conference in Bengaluru, India, formed the centerpiece with additional Developer Days, executive roundtables, user groups, cloud clubs, and specialized workshops. Thank you to all the developers who joined us on this incredible journey across Asia!

AWS Lambda

AWS Lambda now offers console to IDE integration and remote debugging capabilities that streamline the developer workflow from browser to Visual Studio Code and its clones. These enhancements reduce context switching and help developers debug Lambda functions directly in their preferred IDE environment.

The console to IDE feature provides an Open in Visual Studio Code button, enabling developers to move quickly from viewing their function in the browser to editing it in their IDE. AWS automatically handles setup, including the AWS Toolkit installation. Developers can also install dependencies and make code changes, which can automatically sync back to the cloud console. Watch the video to see how it works:

Remote debugging allows you to reduce debugging time from hours to minutes while simplifying local environment setups. You can set breakpoints and debug Lambda functions running in the actual cloud environment with complete access to Amazon VPC resources and AWS Identity and Access Management (AWS IAM) execution roles. The debugging connection uses AWS IoT Secure Tunneling Service, and AWS automatically cleans up debugging configuration after completion. Watch the video to see how it works:

Lambda now integrates with LocalStack directly in the AWS Toolkit for Visual Studio Code. This simplifies local testing of serverless applications involving multiple AWS services. You can now deploy serverless applications to LocalStack using the same commands, debug Lambda functions with one-click setup, and test end-to-end event-driven workflows locally before deploying to the cloud.

Lambda response streaming now supports a maximum response payload size of 200 MB, 10 times higher than before. Response streaming helps build applications that progressively stream response payloads back to clients, improving performance for latency-sensitive workloads by reducing time to first byte (TTFB) performance.

AWS Lambda Hackathon

The AWS Lambda Hackathon, challenged developers to build serverless applications solving real-world business problems using Lambda. With 3,732 participants and 331 project submissions, the competition showcased innovative serverless solutions across diverse domains.

Figure 2: AWS Lambda Hackathon

Figure 2: AWS Lambda Hackathon

We announced winners on July 22, 2025, with $15,000 in total prizes awarded:

  • First Place ($6,000): ForestShield: AWS Deforestation Detection by Younes Laaroussi is a serverless forest monitoring system that tracks deforestation in real-time.
  • Second Place ($4,000): Smart Meeting Assistant by Eduard-David Jitareanu lets you upload audio recordings to create and manage Jira tasks automatically.
  • Third Place ($3,000): Drone SoundAware by Ian Brumby allows drone operators to plan, assess, and optimize flight routes while reducing noise impact on communities.
  • Honorable Mentions ($500):
    • OutScan by Sheldon Aristide is an AI-powered, serverless genomic radar analyzing viral mutations in real-time to detect pandemic threats.
    • Buzz CSV by Damien Pace transforms Excel files into actionable insights through natural language queries.
    • Smart Clip AI by Alexander Bolaño turns long videos into short, high-impact clips.
    • VA Rating Assistant by Chris Lassiter helps you upload medical documents and uses AI to identify potential VA disability claims and ratings, helping veterans access benefits faster and more accurately.

Amazon ECS

Amazon ECS now offers Managed Instances, a new compute option that combines EC2 flexibility with fully managed infrastructure. The functionality automatically handles instance provisioning, scaling, and maintenance while allowing you to use the full range of EC2 capabilities. Key features include:

  • Automated security patching every 14 days with configurable maintenance windows
  • Intelligent task placement and resource optimization across instances
  • Support for custom instance attributes including GPU, CPU architecture, and network performance requirements
  • Built on Bottlerocket OS with automated security updates
  • Deep integration with EC2 pricing options
  • Default cost-optimized instance selection with option for custom specifications

Watch the video to learn more.

Amazon ECS now enables built-in blue/green deployments. This reduces the need for custom deployment tooling while making containerized application releases safer and more reliable. The new capability provisions the new application version (green) alongside the existing version (blue), allowing validation before routing production traffic. ECS also introduced deployment lifecycle hooks powered by Lambda functions that integrate custom validation steps multiple stages of deployment. Watch the video to learn more.

Amazon S3

Amazon S3 Vectors is now available in preview. This is a cloud object store with native support for storing vector datasets and with sub-second query performance for AI applications. Vector buckets is a new bucket type with dedicated APIs for storing, accessing, and querying vector data without infrastructure provisioning.

Figure 3: Amazon S3 Vectors

Figure 3: Amazon S3 Vectors

Amazon S3 Metadata now supports metadata for all your S3 objects. This allows you to analyze and query metadata for your entire S3 storage footprint. S3 Metadata live inventory tables gives you a fully managed Apache Iceberg table, including existing objects. This provides a fully managed snapshot of all objects and metadata, refreshed within 1 hour of changes. S3 Metadata journal tables offer a near-real-time view of object-level changes.

S3 also now supports a preview in the AWS Console for S3 Tables, making it easier to understand data structure and content without writing SQL. S3 Batch Operations now supports bulk target selection for managing buckets through the console. S3 also now supports conditional deletes in S3 general purpose buckets, allowing safer deletion operations.

Amazon EventBridge

Amazon EventBridge now provides enhanced logging capabilities with detailed information about successes, failures, and status codes. This new observability feature provides visibility into the complete event journey, showing when events are published, matched against rules, delivered to subscribers, or encounter failures. You can send logs to Amazon CloudWatch Logs, S3, or Amazon Data Firehose.

Generative AI with serverless

Discover how to effectively build AI agents on AWS Serverless shows how to use Amazon Bedrock AgentCore, Lambda, and ECS to build production-ready agentic AI systems. The blog explains how to use the Strands Agents SDK, which is a framework for building AI agents. The post includes storing session state, implementing authentication using Amazon Cognito and Amazon API Gateway, integrating tools through MCP, and establishing observability using OpenTelemetry.

Figure 4: Agentic loop

Figure 4: Agentic loop

A series on serverless generative AI architectural patterns (part1, part2) explores non-real-time generative AI scenarios. These include buffered asynchronous request-response using Amazon SQS queues, multimodal parallel fan-out using EventBridge or Amazon SNS, and non-interactive batch processing using AWS Step Functions or AWS Glue.

Kiro: Spec-driven AI development

AWS introduced Kiro, an agentic AI-powered IDE now available in preview. It is built on the open-source Code OSS platform (the same foundation as VS Code) so you can use your existing extensions. Kiro brings a spec-driven approach to software development that bridges the gap between rapid prototyping and production-ready code. Kiro emphasizes structured development. It breaks down developer prompts into comprehensive requirements, system design documents, and task lists before writing any code. You can download Kiro for macOS, Windows, and Linux from the Kiro website.

Amazon Bedrock AgentCore

Amazon Bedrock AgentCore is now available in preview, offering a set of services that help developers quickly and securely deploy AI agents at scale. AgentCore supports frameworks including CrewAI, LangGraph, LlamaIndex, and Strands Agents, and works with any model in or outside Amazon Bedrock.

AgentCore includes seven modular services:

  • AgentCore Runtime provides sandboxed low-latency serverless environments with up to 8-hour runtime support
  • AgentCore Memory manages both short-term and long-term memory with built-in policies
  • AgentCore Observability offers step-by-step visualization with OpenTelemetry support
  • AgentCore Identity provides a secure token vault for OAuth 2.0 and API keys.
  • AgentCore Gateway transforms APIs and Lambda functions into agent-ready tools with a unified MCP interface
  • AgentCore Browser enables managed web automation
  • AgentCore Code Interpreter provides safe code execution environments.

Amazon Bedrock

Amazon Bedrock continues to expand its foundation model selection with new models now generally available. Qwen models bring four fully managed open-weight models which excel at sophisticated coding tasks, multi-tool agentic workflows, and adaptive reasoning through hybrid thinking modes. DeepSeek-V3.1 delivers performance improvements on certain benchmarks while maintaining cost efficiency through its mixture-of-experts architecture.

Amazon SNS

Amazon SNS now supports three additional message filtering operators: wildcard matching, anything-but wildcard matching, and anything-but prefix matching. SNS also now supports message group IDs in standard topics, enabling fair queue functionality for subscribed SQS standard queues.

Serverless Compute Blog Posts

July

August

September

Serverless Office Hours weekly livestream

July

August

September

Videos

The Serverless landing page has more information. The Lambda resources page contains case studies, webinars, whitepapers, customer stories, reference architectures, and even more Getting Started tutorials.

You can also follow the Developer Advocacy team members who work on Serverless to see the latest news, follow conversations, and interact with the team.

And finally, visit ServerlessLand for all your serverless needs.