Bedrock Brief 15 Oct 2025

Pleri AI

Pleri AI

— 4 min read
Bedrock Brief 15 Oct 2025

Hold onto your cloud hats, folks! This week in AWS AI land, we've got more shakeups than a bartender at happy hour.

First up, AWS is beefing up its security muscle faster than you can say "ChatGPT data breach." They've snagged Chet Kapoor, former DataStax CEO, as the new VP of security services and observability. Reporting directly to AWS CEO Matt Garman, Kapoor's mission is clear: fortify those AI castle walls. It's like AWS is playing chess while the competition's still figuring out how to set up the board. Read more about the security shuffle here.

Meanwhile, Amazon's dropping automation tools like they're hot. Enter Amazon Quick Suite, the new kid on the block promising to turn your mundane tasks into a set-it-and-forget-it dream. Picture this: a program manager at AWS used it to automate their weekly report, saving hours of mind-numbing work. It's like having a digital intern who never complains about coffee runs.

But wait, there's more! Oracle's playing nice in the AWS sandbox, expanding Oracle Database@AWS with AI goodies. They've added support for Oracle Autonomous AI Lakehouse and other fancy features. It's like watching two frenemies reluctantly share their toys in the cloud playground. The real winners? Enterprises who can now have their Oracle cake and eat it on AWS too.

Fresh Cut

  • EC2 M7i instances, featuring custom Intel processors with 15% better performance than competitors, are now available in Milan, offering larger sizes and bare metal options with built-in accelerators for enhanced data processing. Read announcement →
  • Amazon QuickSight users can now customize fonts for data labels and axes in charts, enhancing dashboard readability and brand consistency. Read announcement →
  • SageMaker AI Projects allows administrators to store custom ML project templates in S3, enabling data scientists to create standardized projects that align with organizational requirements directly from SageMaker AI studio. Read announcement →
  • Amazon Bedrock AgentCore lets developers build and run AI agents that can interact with tools and data for up to eight hours, without managing infrastructure, using any framework or model. Read announcement →
  • Amazon ElastiCache now offers vector search capabilities, allowing developers to efficiently index and retrieve high-dimensional vector embeddings with microsecond latency, enabling faster and more cost-effective AI applications like semantic caching and retrieval augmented generation. Read announcement →
  • Amazon CloudWatch's new generative AI observability feature gives developers a comprehensive view of AI application performance, including token usage, latency, and errors across all components, helping to quickly identify and troubleshoot issues in AI workloads. Read announcement →
  • Amazon's new Quick Suite integrates AI agents with your business data, enabling seamless information retrieval and task automation across multiple applications and databases, potentially revolutionizing how developers and teams interact with their data and workflows. Read announcement →
  • Amazon SageMaker notebook instances can now use Amazon Linux 2023, offering developers a more secure and up-to-date environment for machine learning projects with features like SELinux support and FIPS 140-3 validated cryptographic modules. Read announcement →
  • Amazon Q Developer now answers pricing questions and estimates workload costs, helping developers make informed decisions about AWS resources without sifting through multiple pricing pages. Read announcement →

The Quarry

Medical reports analysis dashboard using Amazon Bedrock, LangChain, and Streamlit

Picture this: a medical dashboard that's less "confusing spreadsheet" and more "AI-powered crystal ball." By teaming up Amazon Bedrock's language models with LangChain's document wizardry and Streamlit's eye-candy visuals, developers can whip up a system that turns dense medical reports into bite-sized insights faster than you can say "stat." The secret sauce? A context-aware chat system that doesn't just regurgitate facts, but actually understands the nuances of medical jargon, making it a powerful sidekick for healthcare professionals trying to decode patient data. Read blog →

More posts:

Core Sample

Text Classification: Handling 1000+ Classes with Generative AI

Lior Perez showcases a nifty three-step text classification workflow that can handle over 1000 categories using Amazon Bedrock, perfect for those times when your data feels like it's playing a game of "category hide-and-seek." The secret sauce? A clever combo of semantic search, reranking, and attribute validation that keeps accuracy high without making your CPU cry. For the code-curious, there's a GitHub repo waiting to spill all the technical tea on how to implement this classification sorcery in your own projects. Watch video →

More videos:

Read more