Blog_dumb

Amazon Bedrock Guardrails now supports protection against undesirable content within code elements including user prompts, comments, variables, function names, and string literals. Amazon Bedrock Guardrails provides configurable safeguards for building generative AI applications at scale. These safety controls work seamlessly whether you’re using foundation models from Amazon Bedrock, or applying them at various intervention points …

Amazon Bedrock Guardrails expands support for code domain Read More »

As AI applications grow more complex, many builders struggle to appropriately and responsibly balance AI benefits and risks. Few resources exist that help non-experts articulate and resolve the key design decisions they must make. However, it doesn’t have to be this way. Today, we’re announcing the AWS Well-Architected Responsible AI Lens—a set of thoughtful questions …

Announcing the AWS Well-Architected Responsible AI Lens  Read More »

At Amazon, we screen customers and transactions across our global business and its subsidiaries to comply with sanctions and other global laws. Failure to comply with these laws can result in severe financial penalties and reputational harm. Amazon’s Compliance team has developed an AI-driven screening and investigations system that has transformed Amazon’s compliance processes into …

How Amazon uses AI agents to support compliance screening of billions of transactions per day Read More »

This post was written with Bharath Suresh and Mary Law from Snowflake. Agentic AI is a type of AI that functions autonomously, automating a broader range of tasks with minimal supervision. It combines traditional AI and generative AI capabilities to make decisions, perform tasks, and adapt to its environment without constant human intervention. These autonomous …

Build an agentic solution with Amazon Nova, Snowflake, and LangGraph Read More »

Optimizing generative AI applications relies on tailoring foundation models (FMs) using techniques such as prompt engineering, RAG, continued pre-training, and fine-tuning. Efficient fine-tuning is achieved by strategically managing hardware, training time, data volume, and model quality to reduce resource demands and maximize value. Spectrum is a new approach designed to pinpoint the most informative layers …

Using Spectrum fine-tuning to improve FM training efficiency on Amazon SageMaker AI Read More »

From AI education to disaster response, see how collaboration is at the heart of the work at Google Research. ​ 

Large language models (LLMs) now support a wide range of use cases, from content summarization to the ability to reason about complex tasks. One exciting new topic is taking generative AI to the physical world by applying it to robotics and physical hardware. Inspired by this, we developed a game for the AWS re:Invent 2024 …

Bringing tic-tac-toe to life with AWS AI services Read More »

Amazon SageMaker HyperPod is a purpose-built infrastructure for optimizing foundation model training and inference at scale. SageMaker HyperPod removes the undifferentiated heavy lifting involved in building and optimizing machine learning (ML) infrastructure for training foundation models (FMs). As AI moves towards deployment adopting to a multitude of domains and use cases, the need for security …

HyperPod enhances ML infrastructure with security and storage Read More »

Over the past two years, I’ve worked with many customers using generative AI to transform their organizations. Most stall at experimentation, because costs stack up and timelines extend before delivering demonstrable value. A 2023 AWS MIT Chief Data Officer (CDO) Symposium survey backs this up, reporting that while 71% of Chief Data Officers were experimenting …

Accelerating generative AI applications with a platform engineering approach Read More »

Learn more about Gemini 3 and how it’s upgrading Google Search and AI Mode. ​