TL;DR Using Google’s new Veo 3.1 video model, we created a breathtaking 1 minute 40 second FPV drone flight through mountain valleys, and it took just 15 minutes to generate. Imagine soaring through alpine valleys, gliding between snowy peaks, and diving toward rivers that twist like silver ribbons below, all without leaving your desk. That’s …
Conditional generative modeling aims to learn a conditional data distribution from samples containing data-condition pairs. For this, diffusion and flow-based methods have attained compelling results. These methods use a learned (flow) model to transport an initial standard Gaussian noise that ignores the condition to the conditional data distribution. The model is hence required to learn …
For decades, SQL has been the universal language for data analysis, offering access to analytics on structured data. Large Language Models (LLMs) like Gemini now provide a path to get nuanced insights from unstructured data such as text, image and video. However, integrating LLMs into standard SQL flow requires data movement, at least some prompt …
Read more “Announcing BigQuery-managed AI functions for better SQL”
We recently announced the support for Agent-to-Agent (A2A) protocol on Amazon Bedrock AgentCore Runtime. With this addition, agents can discover peers, share capabilities, and coordinate actions across platforms using standardized communication. Amazon Bedrock AgentCore Runtime provides a secure, serverless environment designed for deploying AI agents and tools. It works with any framework and model, supports …
Read more “Introducing agent-to-agent protocol support in Amazon Bedrock AgentCore Runtime”
Embeddings are a crucial component at the intersection of data and AI. As data structures, they encode the inherent meaning of the data they represent, and their significance becomes apparent when they are compared to one another. Vector search is a technique that uncovers the relative meaning of those embeddings by evaluating the distances between …
Read more “BigQuery under the hood: How Google brought embeddings to analytics”
Extracting structured data from documents like invoices, receipts, and forms is a persistent business challenge. Variations in format, layout, language, and vendor make standardization difficult, and manual data entry is slow, error-prone, and unscalable. Traditional optical character recognition (OCR) and rule-based systems often fall short in handling this complexity. For instance, a regional bank might …
Read more “Fine-tune VLMs for multipage document-to-JSON with SageMaker AI and SWIFT”
As Large Language Models (LLMs) evolve, Reinforcement Learning (RL) is becoming the crucial technique for aligning powerful models with human preferences and complex task objectives. However, enterprises that need to implement and scale RL for LLMs are facing infrastructure challenges. The primary hurdles include the memory contention from concurrently hosting multiple large models (such as …
Read more “Running high-scale reinforcement learning (RL) for LLMs on GKE”
This paper was accepted at the Workshop on Unifying Representations in Neural Models (UniReps) at NeurIPS 2025. Activation steering methods in large language models (LLMs) have emerged as an effective way to perform targeted updates to enhance generated language without requiring large amounts of adaptation data. We ask whether the features discovered by activation steering …
Read more “ExpertLens: Activation Steering Features Are Highly Interpretable”
Organizations need seamless access to their structured data repositories to power intelligent AI agents. However, when these resources span multiple AWS accounts integration challenges can arise. This post explores a practical solution for connecting Amazon Bedrock agents to knowledge bases in Amazon Redshift clusters residing in different AWS accounts. The challenge Organizations that build AI …
Read more “Connect Amazon Bedrock agents to cross-account knowledge bases”
n8n is a powerful yet easy-to-use workflow and automation tool for multi-step AI agents, and many teams want a simple, scalable, and cost-effective way to self-host it. With just a few commands, you can deploy n8n to Cloud Run and have it up and running, ready to supercharge your business with AI workflows that can …
Read more “Easy AI workflow automation: Deploy n8n on Cloud Run”