In this series of posts, you will learn how to build intelligent AI voice agents using Pipecat, an open-source framework for voice and multimodal conversational AI agents, with foundation models on Amazon Bedrock. It includes high-level reference architectures, best practices and code samples to guide your implementation.  ( 8 min )

In this post, we explore the implementation details of a web application that uses the browser’s Web Audio API and Amazon Transcribe streaming to enable real-time dual-channel transcription. By using the combination of AudioContext, ChannelMergerNode, and AudioWorklet, we were able to seamlessly process and encode the audio data from two microphones before sending it to Amazon Transcribe for transcription.  ( 8 min )

At Kepler, a global full-service digital marketing agency serving Fortune 500 brands, we understand the delicate balance between creative marketing strategies and data-driven precision. In this post, we share how implementing Amazon Q Business transformed our operations by democratizing AI access across our organization while maintaining stringent security standards, resulting in an average savings of 2.7 hours per week per employee in manual work and improved client service delivery.  ( 7 min )

TL;DR PyPI provides a neat way of distributing binaries from other languages, and Python venvs make it easy to run different versions side by side. This post takes a look at how to do that with Dart, and the next steps necessary to do a proper job of it. Background A few days ago I […]  ( 14 min )

View on sreweekly.com A message from our sponsor, PagerDuty: 🔍 Notable PagerDuty shift: Full incident management now spans all paid tiers. The upgraded Slack-first and Teams-first experience means fewer tools to juggle during incidents. Only leveraging PagerDuty for basic alerting? Time to check out what’s newly available in your plan! https://fnf.dev/4dZ5V36 You can’t prevent your […]  ( 4 min )

In this post, we demonstrate how to use the Open AI Whisper foundation model (FM) Whisper Large V3 Turbo, available in Amazon Bedrock Marketplace, which offers access to over 140 models through a dedicated offering, to produce near real-time transcription. These transcriptions are then processed by Amazon Bedrock for summarization and redaction of sensitive information.  ( 9 min )

In this post, we demonstrate how to use large vision models (LVMs) for semantic video search using natural language and image queries. We introduce some use case-specific methods, such as temporal frame smoothing and clustering, to enhance the video search performance. Furthermore, we demonstrate the end-to-end functionality of this approach by using both asynchronous and real-time hosting options on Amazon SageMaker AI to perform video, image, and text processing using publicly available LVMs on the Hugging Face Model Hub. Finally, we use Amazon OpenSearch Serverless with its vector engine for low-latency semantic video search.  ( 14 min )

In this post, we discuss how an enterprise with multiple accounts can access a shared Amazon SageMaker HyperPod cluster for running their heterogenous workloads. We use SageMaker HyperPod task governance to enable this feature.  ( 9 min )

This post evaluates the key options for querying data using generative AI, discusses their strengths and limitations, and demonstrates why Text-to-SQL is the best choice for deterministic, schema-specific tasks. We show how to effectively use Text-to-SQL using Amazon Nova, a foundation model (FM) available in Amazon Bedrock, to derive precise and reliable answers from your data.  ( 10 min )

TL;DR Start writing down why decisions are made. Future you may thank you. Future other person who’s wondering what you were thinking may also thank you. Then keep a dependency graph of the things impacted by the decision. It will help unravel what gets woven around it. Background I was at an excellent AFCEA event […]  ( 14 min )

Radial is the largest 3PL fulfillment provider, also offering integrated payment, fraud detection, and omnichannel solutions to mid-market and enterprise brands. In this post, we share how Radial optimized the cost and performance of their fraud detection machine learning (ML) applications by modernizing their ML workflow using Amazon SageMaker.  ( 15 min )

Contextual retrieval enhances traditional RAG by adding chunk-specific explanatory context to each chunk before generating embeddings. This approach enriches the vector representation with relevant contextual information, enabling more accurate retrieval of semantically related content when responding to user queries. In this post, we demonstrate how to use contextual retrieval with Anthropic and Amazon Bedrock Knowledge Bases.  ( 11 min )

In this post, we demonstrate how to deploy a small language model on SageMaker AI by extending our pre-built containers to be compatible with AWS Graviton instances. We first provide an overview of the solution, and then provide detailed implementation steps to help you get started. You can find the example notebook in the GitHub repo.  ( 11 min )

In this post, we share how Impel enhances the automotive dealership customer experience with fine-tuned LLMs on SageMaker.  ( 8 min )

In this post, we show how climate tech startups are developing foundation models (FMs) that use extensive environmental datasets to tackle issues such as carbon capture, carbon-negative fuels, new materials design for microplastics destruction, and ecosystem preservation. These specialized models require advanced computational capabilities to process and analyze vast amounts of data effectively.  ( 13 min )

In this post, we'll explore how to combine Amazon Bedrock prompt caching with Claude Code—a coding agent released by Anthropic that is now generally available. This powerful combination transforms your development workflow by delivering lightning-fast responses from reducing inference response latency, as well as lowering input token costs.  ( 10 min )

We’ve witnessed remarkable advances in model capabilities as generative AI companies have invested in developing their offerings. Language models such as Anthropic’s Claude Opus 4 & Sonnet 4, Amazon Nova, and Amazon Bedrock can reason, write, and generate responses with increasing sophistication. But even as these models grow more powerful, they can only work with […]  ( 16 min )

The Danish humanitarian organization Bevar Ukraine has developed a comprehensive virtual generative AI-powered assistant called Victor, aimed at addressing the pressing needs of Ukrainian refugees integrating into Danish society. This post details our technical implementation using AWS services to create a scalable, multilingual AI assistant system that provides automated assistance while maintaining data security and GDPR compliance.  ( 8 min )

In this post, we explore how Noodoe uses AI and Amazon Bedrock to optimize EV charging operations. By integrating LLMs, Noodoe enhances station diagnostics, enables dynamic pricing, and delivers multilingual support. These innovations reduce downtime, maximize efficiency, and improve sustainability. Read on to discover how AI is transforming EV charging management.  ( 8 min )

Sometimes I need an older or newer version of CMake to the one installed by the system package manager on whatever I’m using, and I’ve found using a Python venv provides an easy way to do that. It’s all facilitated by the fact that CMake is a PyPI package [1]. For example, my Kubuntu desktop […]  ( 13 min )

In this post, we explore how to use Graph-based Retrieval-Augmented Generation (GraphRAG) in Amazon Bedrock Knowledge Bases to build intelligent applications. Unlike traditional vector search, which retrieves documents based on similarity scores, knowledge graphs encode relationships between entities, allowing large language models (LLMs) to retrieve information with context-aware reasoning.  ( 11 min )

This blog post presents an MLOps solution that uses AWS Cloud Development Kit (AWS CDK) and services like AWS Step Functions, Amazon EventBridge and Amazon Personalize to automate provisioning resources for data preparation, model training, deployment, and monitoring for Amazon Personalize.  ( 14 min )

When a regulatory body like the US Food and Drug Administration (FDA) introduces changes to regulations, organizations are required to evaluate the changes against their internal SOPs. When necessary, they must update their SOPs to align with the regulation changes and maintain compliance. In this post, we show different approaches using Amazon Bedrock to identify relationships between regulation changes and SOPs.  ( 18 min )

Pupdate The fair weather has (mostly) continued, which allowed for some nice long walks. Milo turned four at the start of the month :) Brussels The end of the month brought the half term holiday, and Mrs S wanted to spend the first weekend away somewhere. Brussels quickly made the top of the list after […]  ( 14 min )

Pupdate The fair weather has (mostly) continued, which allowed for some nice long walks. Milo turned four at the start of the month :) Brussels The end of the month brought the half term holiday, and Mrs S wanted to spend the first weekend away somewhere. Brussels quickly made the top of the list after […]  ( 14 min )

View on sreweekly.com Automatic rollbacks are a last resort Rollbacks don’t always return you to a previous system state. They can return you to a state you’ve never tested or operated before.   Steve Fenton — Octopus Deploy Burn rate is a better error rate This article explains the math of burn rate alerting and gives […]  ( 3 min )

In this post you define, deploy, and provision a SageMaker Project custom template purely in Terraform. With no dependencies on other IaC tools, you can now enable SageMaker Projects strictly within your Terraform Enterprise infrastructure.  ( 5 min )

ZURU collaborated with AWS Generative AI Innovation Center and AWS Professional Services to implement a more accurate text-to-floor plan generator using generative AI. In this post, we show you why a solution using a large language model (LLM) was chosen. We explore how model selection, prompt engineering, and fine-tuning can be used to improve results.  ( 11 min )

Non-conversational applications offer unique advantages such as higher latency tolerance, batch processing, and caching, but their autonomous nature requires stronger guardrails and exhaustive quality assurance compared to conversational applications, which benefit from real-time user feedback and supervision. This post examines four diverse Amazon.com examples of such generative AI applications.  ( 15 min )

In this post, we give an overview of a well-established generative AI foundation, dive into its components, and present an end-to-end perspective. We look at different operating models and explore how such a foundation can operate within those boundaries. Lastly, we present a maturity model that helps enterprises assess their evolution path.  ( 14 min )

OpenSearch offers a wide range of third-party machine learning (ML) connectors to support this augmentation. This post highlights two of these third-party ML connectors. The first connector we demonstrate is the Amazon Comprehend connector. In this post, we show you how to use this connector to invoke the LangDetect API to detect the languages of ingested documents. The second connector we demonstrate is the Amazon Bedrock connector to invoke the Amazon Titan Text Embeddings v2 model so that you can create embeddings from ingested documents and perform semantic search.  ( 13 min )

Amazon Bedrock Model Copy and Model Share features provide a powerful option for managing the lifecycle of an AI application from development to production. In this comprehensive blog post, we'll dive deep into the Model Share and Model Copy features, exploring their functionalities, benefits, and practical applications in a typical development-to-production scenario.  ( 10 min )

TL;DR I was foolish to believe that paying a premium for a battery with a 5y guarantee would actually get me a battery that lasted 5y :( Background The original battery in my 2010 XC60 died after a little over 5y. I replaced it with an Exide EA852 from Tayna that lasted a bit over […]  ( 14 min )

TL;DR I was foolish to believe that paying a premium for a battery with a 5y guarantee would actually get me a battery that lasted 5y :( Background The original battery in my 2010 XC60 died after a little over 5y. I replaced it with an Exide EA852 from Tayna that lasted a bit over […]  ( 14 min )

In this post, we explore how a leading GeoFM (Clay Foundation’s Clay foundation model available on Hugging Face) can be deployed for large-scale inference and fine-tuning on Amazon SageMaker.  ( 16 min )

In this post, we demonstrate an example of building an agentic RAG application using the LlamaIndex framework. LlamaIndex is a framework that connects FMs with external data sources. It helps ingest, structure, and retrieve information from databases, APIs, PDFs, and more, enabling the agent and RAG for AI applications. This application serves as a research tool, using the Mistral Large 2 FM on Amazon Bedrock generate responses for the agent flow.  ( 13 min )

In this post, we focus on the Amazon Nova Canvas image generation model. We then provide an overview of the image generation process (diffusion) and dive deep into the input parameters for text-to-image generation with Amazon Nova Canvas.  ( 9 min )

In this post, we explore how Amazon Nova Canvas can solve real-world business challenges through advanced image generation techniques. We focus on two specific use cases that demonstrate the power and flexibility of this technology: interior design and product photography.  ( 9 min )

In this post, we walk through how to build a multi-agent investment research assistant using the multi-agent collaboration capability of Amazon Bedrock. Our solution demonstrates how a team of specialized AI agents can work together to analyze financial news, evaluate stock performance, optimize portfolio allocations, and deliver comprehensive investment insights—all orchestrated through a unified, natural language interface.  ( 13 min )

This post explores deploying a text-to-SQL pipeline using generative AI models and Amazon Bedrock to ask natural language questions to a genomics database. We demonstrate how to implement an AI assistant web interface with AWS Amplify and explain the prompt engineering strategies adopted to generate the SQL queries. Finally, we present instructions to deploy the service in your own AWS account.  ( 12 min )

We are excited to announce the availability of Gemma 3 27B Instruct models through Amazon Bedrock Marketplace and Amazon SageMaker JumpStart. In this post, we show you how to get started with Gemma 3 27B Instruct on both Amazon Bedrock Marketplace and SageMaker JumpStart, and how to use the model’s powerful instruction-following capabilities in your applications.  ( 11 min )

In this post, we walk through building a full-stack application that processes multimodal content using Amazon Bedrock Data Automation, stores the extracted information in an Amazon Bedrock knowledge base, and enables natural language querying through a RAG-based Q&A interface.  ( 10 min )

In this post, we show you how to implement and evaluate three powerful techniques for tailoring FMs to your business needs: RAG, fine-tuning, and a hybrid approach combining both methods. We provid ready-to-use code to help you experiment with these approaches and make informed decisions based on your specific use case and dataset.  ( 10 min )

Rufus, an AI-powered shopping assistant, relies on many components to deliver its customer experience including a foundation LLM (for response generation) and a query planner (QP) model for query classification and retrieval enhancement. This post focuses on how the QP model used draft centric speculative decoding (SD)—also called parallel decoding—with AWS AI chips to meet the demands of Prime Day. By combining parallel decoding with AWS Trainium and Inferentia chips, Rufus achieved two times faster response times, a 50% reduction in inference costs, and seamless scalability during peak traffic.  ( 8 min )

Amazon Bedrock Data Automation helps organizations streamline development and boost efficiency through customizable, multimodal analytics. It eliminates the heavy lifting of unstructured content processing at scale, whether for video or audio. The new capabilities make it faster to extract tailored, generative AI-powered insights like scene summaries, key topics, and customer intents from video and audio. This unlocks the value of unstructured content for use cases such as improving sales productivity and enhancing customer experience.  ( 6 min )

In this post, we share how GuardianGamer uses AWS services including Amazon Nova and Amazon Bedrock to deliver a scalable and efficient supervision platform. The team uses Amazon Nova for intelligent narrative generation to provide parents with meaningful insights into their children’s gaming activities and social interactions, while maintaining a non-intrusive approach to monitoring.  ( 7 min )

View on sreweekly.com Security and SRE: How Datadog’s combined approach aims to tackle security and reliability challenges Datadog has fully merged their SRE and Security teams. In this post, we’ll look at essential elements of SRE and security, the benefits we’ve realized by combining the two disciplines, and what that approach looks like for us. […]  ( 4 min )

Public Ollama Models 20250525 How to chat with Ollama models Select an IP and model from the table below, then use them in this command: # Start a conversation with a model # Replace <IP> with an IP from the table below # Replace <MODEL> with one of the models listed for that IP curl -X POST http://<IP>:11434/api/chat -d '{ "model": "<MODEL>", "messages": [{ "role": "user", "content": "Hello, how are you?" }] }' Available Models IP Models 123.60.64.205 llama3.2:3b-instruct-q5_K_M,deepseek-r1:1.5b 101.132.74.29 llama3.2:3b-instruct-q5_K_M 139.196.196.43 qwen3:32b,qwen3:30b-a3b,qwen2.5:72b-instruct,deepseek-r1:70b 223.166.61.66 qwen3:4b,qwen3:32b,mxbai-embed-large:latest,bge-m3:latest,nomic-embed-text:latest,qwen2.5-coder:32b,qwen2.5-coder:14b,qwen2.5:32b,qwen2.5:latest,mistral-small3.1:latest,codellama:13b,codellama:34b,codellama:70b,qwq:latest,deepseek-v2:16b,deepseek-coder-v2:latest,gemma3:12b,gemma3:latest,qwen2.5:3b,gemma3:1b,deepseek-r1:14b,deepseek-r1:32b,deepseek-r1:8b,deepseek-r1:7b,deepseek-r1:1.5b 106.75.235.214 llama3.1:8b,gemma3:27b,qwq:latest,deepseek-r1:70b 117.50.171.129 qwq:latest,qwen2.5-coder:32b 117.50.180.35 bge-m3:latest 116.232.26.17 qwen3:0.6b,qwen3:1.7b 117.50.176.34 qwen2.5:1.5b,qwen2.5:7b,qwen2.5:14b 222.64.131.253 llama3.2:3b-instruct-q5_K_M,qwen2.5:32b-instruct-q4_K_M,qwq:32b-q4_K_M,qwen2.5:32b,qwen2.5-coder:14b,sqlcoder:15b,deepseek-r1:14b,qwen2.5:14b,bge-m3:567m,qwen2.5:7b,nomic-embed-text:latest,qwen2.5:0.5b 123.60.5.218 llama3-backup:latest,mario:latest,llama3.2:latest,llama3.2:3b-instruct-q5_K_M,deepseek-r1:8b 218.1.223.134 llama3.2:3b-instruct-q5_K_M,smollm2:135m,deepseek-r1:1.5b,deepseek-r1:7b,deepseek-r1:8b 101.132.88.19 nomic-embed-text:latest,llama3.2:3b,llama3.2:1b,deepseek-r1:1.5b 114.95.209.211 nomic-embed-text:latest,qwen2.5-coder:latest,deepseek-r1:7b Disclaimer These Ollama model endpoints are publicly exposed interfaces found on the internet. They are listed here for informational purposes only. Please be aware that: These endpoints are not maintained or controlled by us The availability and stability of these services cannot be guaranteed Use these services at your own risk We take no responsibility for any issues or damages that may arise from using these endpoints 免责声明 本文列出的 Ollama 模型接口均来自互联网上公开暴露的端点。请注意： 这些端点并非由我们维护或控制 无法保证这些服务的可用性和稳定性 使用这些服务需自行承担风险 对于使用这些端点可能产生的任何问题或损失，我们不承担任何责任  ( 2 min )

In this post, we explore how Principal used this opportunity to build an integrated voice VA reporting and analytics solution using an Amazon QuickSight dashboard.  ( 7 min )

This post demonstrates how Amazon Bedrock, combined with a user feedback dataset and few-shot prompting, can refine responses for higher user satisfaction. By using Amazon Titan Text Embeddings v2, we demonstrate a statistically significant improvement in response quality, making it a valuable tool for applications seeking accurate and personalized responses.  ( 12 min )

Amazon Q Business integration with Microsoft 365 applications offers powerful AI assistance directly within the tools that your team already uses daily. In this post, we explore how these integrations for Outlook and Word can transform your workflow.  ( 8 min )

In this post, we present a solution to incorporate Amazon Bedrock Agents in your Slack workspace. We guide you through configuring a Slack workspace, deploying integration components in Amazon Web Services, and using this solution.  ( 11 min )

In this post, we present a solution for securing distributed logging multi-account deployments using Amazon Bedrock and LangChain.  ( 11 min )

In this post, we introduce a multi-agent collaboration pipeline for processing unstructured insurance data using Amazon Bedrock, featuring specialized agents for classification, conversion, and metadata extraction. We demonstrate how this domain-aware approach transforms diverse data formats like claims documents, videos, and audio files into metadata-rich outputs that enable fraud detection, customer 360-degree views, and advanced analytics.  ( 16 min )

In this post, we explore how Onity Group, a financial services company specializing in mortgage servicing and origination, transformed their document processing capabilities using Amazon Bedrock and other AWS services. The solution helped Onity achieve a 50% reduction in document extraction costs while improving overall accuracy by 20% compared to their previous OCR and AI/ML solution.  ( 10 min )

HERE collaborated with the GenAIIC. Our joint mission was to create an intelligent AI coding assistant that could provide explanations and executable code solutions in response to users’ natural language queries. The requirement was to build a scalable system that could translate natural language questions into HTML code with embedded JavaScript, ready for immediate rendering as an interactive map that users can see on screen.  ( 12 min )

View on sreweekly.com Human Error Strikes Again… or Does It? Why don’t we look for the root cause of a successful outcome?   Hamed Silatani — Uptime Labs How we optimized LLM use for cost, quality, and safety to facilitate writing postmortems They took a great deal of care to avoid the potential pitfalls of using […]  ( 4 min )

In this post, we demonstrate how to build a custom plugin with Amazon Q Business for backend integration. This plugin can integrate existing systems, including third-party systems, with little to no development in just weeks and automate critical workflows. Additionally, we show how to safeguard the solution using Amazon Cognito and AWS IAM Identity Center, maintaining the safety and integrity of sensitive data and workflows.  ( 12 min )

This post walks you through how to create a basic hallucination detection system for RAG-based applications. We also weigh the pros and cons of different methods in terms of accuracy, precision, recall, and cost.  ( 13 min )

Pit crews are trained to operate at optimum efficiency, although measuring their performance has been challenging, until now. In this post, we share how Amazon Web Services (AWS) is helping Scuderia Ferrari HP develop more accurate pit stop analysis techniques using machine learning (ML).  ( 6 min )

In this post, we demonstrate how to retrain and quantize a model using SageMaker AI and the SiMa.ai Palette software suite. The goal is to accurately detect individuals in environments where visibility and protective equipment detection are essential for compliance and safety.  ( 14 min )

Building on this foundation of specialized information extraction solutions and using the capabilities of SageMaker HyperPod, we collaborate with APOIDEA Group to explore the use of large vision language models (LVLMs) to further improve table structure recognition performance on banking and financial documents. In this post, we present our work and step-by-step code on fine-tuning the Qwen2-VL-7B-Instruct model using LLaMA-Factory on SageMaker HyperPod.  ( 15 min )

In this post, we share how Qualtrics built an AI platform powered by Amazon SageMaker and Amazon Bedrock.  ( 12 min )

AWS partnered with Vxceed to support their AI strategy, resulting in the development of LimoConnect Q, an innovative ground transportation management solution. Using AWS services including Amazon Bedrock and Lambda, Vxceed successfully built a secure, AI-powered solution that streamlines trip booking and document processing.  ( 9 min )

This post is the first in a series where we will run multiple diffusion transformers on Trainium and Inferentia-powered instances. In this post, we show how you can deploy PixArt-Sigma to Trainium and Inferentia-powered instances.  ( 9 min )

This post is the first in a series where we will run multiple diffusion transformers on Trainium and Inferentia-powered instances. In this post, we show how you can deploy PixArt-Sigma to Trainium and Inferentia-powered instances.  ( 9 min )

In this post, we use the recipes to fine-tune the original DeepSeek-R1 671b parameter model. We demonstrate this through the step-by-step implementation of these recipes using both SageMaker training jobs and SageMaker HyperPod.  ( 15 min )

In this post, we show you how Amazon Q Business can help augment your generative AI needs in all the abovementioned use cases and more by answering questions, providing summaries, generating content, and securely completing tasks based on data and information in your enterprise systems.  ( 12 min )

Generative AI tools have transformed how we work, create, and process information. At Amazon Web Services (AWS), security is our top priority. Therefore, Amazon Bedrock provides comprehensive security controls and best practices to help protect your applications and data. In this post, we explore the security measures and practical strategies provided by Amazon Bedrock Agents to safeguard your AI interactions against indirect prompt injections, making sure that your applications remain both secure and reliable.  ( 11 min )

In this post, we demonstrate a solution using Amazon Elastic Kubernetes Service (EKS) with Amazon Bedrock to build scalable and containerized RAG solutions for your generative AI applications on AWS while bringing your unstructured user file data to Amazon Bedrock in a straightforward, fast, and secure way.  ( 7 min )

Recognizing the transformative benefits of generative AI for enterprises, we at Hexagon’s Asset Lifecycle Intelligence division sought to enhance how users interact with our Enterprise Asset Management (EAM) products. Understanding these advantages, we partnered with AWS to embark on a journey to develop HxGN Alix, an AI-powered digital worker using AWS generative AI services. This blog post explores the strategy, development, and implementation of HxGN Alix, demonstrating how a tailored AI solution can drive efficiency and enhance user satisfaction.  ( 13 min )

In this post, we demonstrate how your organization can reduce the end-to-end burden of resolving regular challenges experienced by your IT support teams—from understanding errors and reviewing diagnoses, remediation steps, and relevant documentation, to opening external support tickets using common third-party services such as Jira.  ( 11 min )

Milo is four today, another milestone worth celebrating :) Last week he had a scan, which followed 8 weeks after the end of his third (modified) CHOP protocol. The scan was clear, so we’ll be back to NDSR at the start of July. It’s all a very similar situation to a year ago. Insurance ManyPets […]  ( 13 min )

Milo is four today, another milestone worth celebrating :) Last week he had a scan, which followed 8 weeks after the end of his third (modified) CHOP protocol. The scan was clear, so we’ll be back to NDSR at the start of July. It’s all a very similar situation to a year ago. Insurance ManyPets […]  ( 13 min )

View on sreweekly.com Automation and The Substitution Myth The myth is: The underlying and often unexamined assumption for the benefits of automation is the notion that computers/machines are better at some tasks, and humans are better at a different, non-overlapping set of tasks. This article lays out several pitfalls to this approach, with references.   Courtney […]  ( 4 min )