What shipped in AI engineering: September 2026
216 posts from 13 sources, grouped by area and ranked by relevance to engineers shipping production AI.
Models & LLMs
Model releases, evaluations and LLM research · 39
- Anthropic's Claude Fable 5.1 and Mythos 5.1 arrive with a 75% cost reduction for Fable cache reads
Anthropic unveiled Claude Fable 5.1 (GA) and Claude Mythos 5.1 (restricted)—the same model with differing safeguards—and announced a 75 % cost reduction for cached‑context reads, enabling far cheaper persistent agents.
- The generative AI customization spectrum: From prompt engineering to custom models on AWS
This article introduces an 8‑step decision framework—the generative AI customization spectrum—for selecting the appropriate level of model customization on AWS, ranging from prompt engineering (Step 1‑2) to Retrieval‑Augmented…
- Anthropic releases Claude Opus 5.5 and OpenAI counters with two cheaper GPT-6 models
Anthropic launched Claude Opus 5.5 with a 20% price reduction, and OpenAI responded minutes later by releasing two GPT‑6 models, Sol and Luna, each priced at half the cost of their predecessors.
- Reduce LLM latency with prefix-aware routing on Amazon SageMaker Inference
Amazon SageMaker Inference introduces prefix‑aware routing, a new strategy that directs all requests beginning with the same prompt prefix to a single instance, keeping the KV cache warm.
- Anthropic debuts Claude Sonnet 5.5 running 30% faster than the previous-generation AI model
Anthropic PBC has released Claude Sonnet 5.5, a mid‑tier large language model that delivers everyday task performance while running more than 30 % faster and at a lower cost than its predecessor.
- Your LLM Has a Curved Space of Paragraphs
The article proposes that within transformer architectures, token indices serve as coordinates while paragraph structure imposes a metric, effectively creating a curved space representation for language models.
Show 33 more in Models & LLMs
- Grok 4.7 is now available on Amazon Bedrock
- Google’s new frontier AI model Gemini 4 Argon goes to cybersecurity defenders first
- Model-agnostic PII detection with LLMs
- AMD acquires world model developer World Labs for $8.2B
- Introducing Claude Sonnet 5.5 on AWS
- Open-source AI developer Mistral closes €3B funding round
- SpaceX launches Grok 4.7 with long-horizon processing, safety upgrades
- Reducing medical claims review time with AI on AWS: The EXL Medical IDP solution
- GraphRAG: A Practitioner's Guide to 6 Advanced Architectural Patterns
- Jev vs. LLMs: When AI Moves from Generation to Decision-Making
- Meta says it has caught up with Anthropic and OpenAI with Muse Spark 1.3, its most powerful AI model yet
- How to Make Your Own JEV Model from an Open LLM
- Language Models for Text Classification: From Bag-of-Words to Jev
- Gemini 4 Argon: our next era of frontier intelligence
- Monitoring Embedding Drift in Production Scikit-LLM Pipelines
- Multilingual Text Classification with Scikit-LLM and Multilingual Embeddings
- Treating Prompt Templates as Hyperparameters in Scikit-LLM GridSearchCV
- Take on your most ambitious work with GPT-6 Astra on Amazon Bedrock
- Microsoft overhauls Copilot with new coding, document editing features
- Anthropic launches Claude Fable 5.1 after inking $35B cloud deal with Lambda
- Advancing AI for biology: Teaching models to design and characterize antibodies
- Anthropic opens AI-powered biology research lab
- When the Correct Answer Is Nothing, What Does Your Pipeline Return?
- Anthropic brings Cowork directly inside Claude’s chat interface
- Introducing SynthID Bio
- Versioning and Tracking Scikit-LLM Experiments
- I Trained a Tiny Network to Compress Data. It Drew a Pentagon.
- ‘Now We Can Know Everything and Do Anything,’ Jensen Huang Says at Dreamforce
- Automating Knowledge Graph Population: Extracting Entities and Triples from Unstructured Text with an LLM
- Introducing Claude Fable 5.1 on AWS
- How Open Science Can Help Researchers Prepare for the Next Pandemic
- Resect launches with $25M to reduce hallucinations in AI models
- Anthropic uses Claude to formalize proof of Fermat’s Last Theorem
Agents & MCP
Agent frameworks, tool use, MCP and orchestration · 67
- Automated agent evaluation with Amazon Bedrock AgentCore and GitHub Actions
This article demonstrates how to build a CI/CD quality gate for Amazon Bedrock AgentCore agents by automatically deploying the agent with role‑based MCP tools via CDK, running on‑demand evaluations through the AgentCore Evaluate API,…
- How HPE Zerto built an agentic troubleshooting system with Amazon Bedrock
HPE Zerto has engineered an on‑premises, agentic troubleshooting system that leverages Amazon Bedrock.
- AWS CloudWatch Omni goes after the hardest question in agentic AI: Why did the agent do that?
AWS CloudWatch Omni introduces observability capabilities aimed at answering “Why did the agent do that?” for agentic AI systems, addressing cases where agents meet latency and error‑free metrics yet produce incorrect actions or stale…
- Agentic AI security pushes vendors toward shared safeguards
Agentic AI security is emerging as a cross‑industry challenge as autonomous systems proliferate across products and platforms.
- From Data to Dialogue: How S&P Global Energy Made Its Structured Data Estate Conversational with Databricks Genie Agents and MCP
S&P Global Energy transformed its structured data estate into a conversational interface by deploying Databricks Genie Agents integrated with the Model Context Protocol (MCP).
- Deploy and manage coding agents at scale with the Unity Gateway CLI
The Unity Gateway CLI provides a command‑line interface for deploying and managing coding agents at scale, supporting the latest large language models such as GPT‑6, Claude Opus 5.5, Gemini 3.8, and Grok 4.7.
Show 61 more in Agents & MCP
- Deploy Hugging Face models on Amazon SageMaker AI with coding agents
- Beyond the price per token: Choosing the right OpenAI model on Amazon Bedrock for your workload
- Optimizing agent system prompts with Amazon Bedrock AgentCore
- Abnormal AI: Amazon Bedrock AgentCore for agentic email security at scale
- Manage end-user OAuth consent for AI agents with Amazon Bedrock AgentCore
- Oracle expands AI agent features with Fusion Claw
- ServiceNow calls for a measured response to rogue AI agents
- Build a multi-account AI agent with AgentCore Gateway and MCP
- AI agent certification startup AIUC raises $40M to begin auditing frontier models
- Agents expose the limitations of trust
- Build interactive MCP Apps using Amazon Bedrock AgentCore
- Implementing defense-in-depth authorization for MCP tools on Amazon Quick
- Agentic AI puts new pressure on identity and database security
- CData’s AI gateway governs agents’ access to enterprise data
- With $35M in early funding, Baselayer is building the trust layer for the agentic economy
- Copado extends Agentia agentic AI DevOps platform for Salesforce with headless automation
- Omnissa debuts AI agents for IT, a managed cloud PC service and Elara for AI governance
- How to scale agentic applications without creating AI sprawl
- Why Most Multi-Agent Systems Fail Even When Evaluation Passes
- Teradata expands AI assistant with tools for governed agents
- On theCUBE Pod: Dreamforce unveils the agentic dream, and Nscale files to go public
- How Ninth Wave built AI-powered open finance onboarding on Amazon Bedrock
- From portal-hopping to instant answers: HEMA’s journey with MCP and Amazon Bedrock
- A shared agentic platform for Wood Mackenzie, on Amazon Bedrock AgentCore
- Extending public sector intelligence with Agentforce and AWS
- Collibra brings runtime governance to enterprise AI agents
- Monitoring production agent lifecycle with AWS DevOps Agent and AgentCore Evaluations
- Genie One MCP: Give any AI Agent the Right Business Context
- Designing lifecycle policies for AgentCore memory
- Best practices for building agentic automations with Amazon Quick Automate
- Agentic conversational video intelligence built on AWS
- NVIDIA Isaac ROS 5.0 Advances Agentic, Open Source Robotics Development
- How Reactiv automates mobile commerce 80% faster with Amazon Bedrock AgentCore
- How Trane gets building insights 60x faster with Amazon Bedrock AgentCore
- Migrate agentic workloads to Amazon Bedrock AgentCore
- From code to diagrams: Agentic architecture documentation with Amazon Bedrock AgentCore
- Trinity: Agentic AI-powered transition planning for students with disabilities
- Ekai raises $1.7M to give enterprise AI agents verified business context
- The Genie One MCP is now Generally Available
- How Intuit built an agentic disaster recovery assistant with Amazon Bedrock
- AI-driven development lifecycle using Amazon Bedrock AgentCore
- Local Agentic AI Workflows with Hermes + Ollama
- Ema raises $77M in funding to deploy AI employees across enterprise HR, IT and finance departments
- Fine-Tuning Agentic AI: A Practical Guide
- From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI
- When One Process Becomes Too Much: Splitting a Pipeline into MCP Services
- Rocket Software brings governed AI agents to mainframe operations
- Geordie launches Cost Intelligence to tie AI spending to agent activity
- From MIT to IBM, expediting AI and quantum deployment
- From theory to delivery: How Atos upskilled 400 engineers in agentic AI
- Multi-Agent Coding Isn’t Enough — Agents Need a Commitment Layer
- Claude Opus 5.5 is now available on AWS
- Amazon makes its agentic AI platform Quick generally available for desktop on Windows and macOS
- Modernizing and scaling support operations with generative AI on AWS
- Equals Money lets customers’ AI tools read data but not move money
- Salesforce introduces Enterprise AI Harness, AI Control Plane
- When agents move at machine speed, security teams lose their lag time
- CrowdStrike builds an identity provider for AI agents, not humans
- At AI Day Singapore, NVIDIA and Partners Showcase AI Advancements Across Southeast Asia
- Nvidia PAIR makes it easy to create a household data center for running agentic AI tasks
- The Pulse: a new trend of CPU shortages
Inference & infrastructure
Serving, latency, GPUs and deployment · 50
- Benchmarking small LLM inference on SageMaker AI: G7 vs G5 and G6
This article benchmarks two 30B Mixture‑of‑Experts models on Amazon SageMaker AI inference across G5, G6, and the newer G7 GPU instance families, showing that the Blackwell‑based G7 (RTX PRO 4500) can achieve superior throughput,…
- Microsoft AI’s MAI-Transcribe-2 undercuts OpenAI, Google and ElevenLabs on price and speed
Microsoft AI unveiled MAI‑Transcribe‑2, a speech‑recognition model priced at 10 ¢ per hour of audio—about a 72 % cut from its prior $0.36‑per‑hour rate—and claims a 5.2 % average word error rate, topping the FLEURS benchmark across 60…
- Your files stay put: Perplexity’s hybrid AI keeps confidential data off the cloud
Perplexity’s new hybrid compute feature for its agentic platform, Computer, lets a single AI agent split its workload between powerful frontier models in the cloud and smaller open‑weight models running locally on Apple silicon Macs.
- Deploying Qwen3.8-2.4T-A95B on Amazon SageMaker HyperPod with vLLM
The authors demonstrate how to deploy the 2.4‑trillion‑parameter Qwen3.8‑2.4T‑A95B model on Amazon SageMaker HyperPod using vLLM, achieving an OpenAI‑compatible endpoint that supports built‑in reasoning, tool‑calling, and native MTP…
- Salesforce debuts Koa, a specialized model built to reason about CRM data
Salesforce announced Koa, a domain‑specific AI model designed to reason over CRM data, built on Nvidia’s Nemotron architecture.
- 4 insights from Dell’s AI Leadership Symposium: Cost and control reshape enterprise AI deployment strategy
Dell’s AI Leadership Symposium highlighted that enterprises are moving past proof‑of‑concepts into production, where cost, data governance, and operational control are becoming the primary drivers of AI deployment strategies.
Show 44 more in Inference & infrastructure
- Introducing Amazon SageMaker HyperPod Inference Gateway
- Set up OpenAI ChatGPT Codex with LiteLLM on Amazon ECS and Amazon Bedrock
- Google launches two benchmark-topping speech generation models
- Run Positron on Amazon SageMaker AI for data science workflows
- CoreWeave expands full-stack AI cloud push as inference demand grows
- Why Deploying Physical AI at Scale Demands Safety at Every Layer
- Deploying real-time personalized speech with Qwen3-TTS on Amazon SageMaker AI
- Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
- Multi-Region training with Amazon SageMaker HyperPod and Qumulo
- Build a serverless PII redaction pipeline with Amazon Bedrock Data Automation
- Govern models with MLflow and Amazon SageMaker AI Model Registry sync: Part 2
- How Diffusion Controller unifies and simplifies AI image generation
- How Tata Elxsi detects industrial safety risks in seconds on AWS
- Pathway’s brain-inspired architecture development on Amazon SageMaker HyperPod
- Aranya raises $11M to turn bare-metal servers into AI clusters in less than 48 hours
- Simplify and support your TorchServe workloads using Ray Serve Deep Learning Containers
- We Pinned Our Model Version to Stay Safe. The Provider Deprecated It Anyway.
- NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut
- Right-size generative AI endpoints with concurrency sweeps on Amazon SageMaker AI
- Speaker-labeled transcription with WhisperX on SageMaker AI
- Amazon SageMaker Inference: 2026 year-to-date launches in review
- d-Matrix Adopts NVIDIA NVLink Fusion for Rack-Scale XPU Deployment
- Fault tolerant distributed training on Amazon EKS using NVRx
- The KV Cache Tax: Why Inference Servers Run Out of Memory Before Compute
- AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories
- Disaggregation Is a Thousand-GPU Problem
- Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX
- Advancing Private AI Compute with secure, server-side memory
- A kernel-centric path to real-time video generation on Trainium
- Sparks Fly: NVIDIA Accelerates Local AI at IFA 2026
- AI drug discovery startup Basecamp Research raises $140M
- NVIDIA Launches DSX Ready to Qualify Power and Cooling Products for AI Factories
- University of Manchester Uses NVIDIA Earth-2 to Forecast Air Pollution Across the UK
- Skild AI Taps NVIDIA Physical AI to Teach Robots New Tasks From a Single Video
- Physical AI Takes the Wheel: How the World’s Robotaxi Leaders Are Building With NVIDIA Technologies
- NVIDIA Opens Applications for 2027–2028 Graduate Fellowships With Awards Up to $60,000
- Sakeena Fiza Helps NVIDIA Hardware Succeed at Scale
- From Enablement to Execution, Egypt’s AI Ecosystem Reaches Production Scale
- 5 Companies Using NVIDIA AI for Clean Energy
- Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data Centers
- ‘NBA 2K27’ With NVIDIA DLSS 5 Leads 28 New Games Coming to GeForce NOW
- NVIDIA to Acquire Hugging Face
- Nvidia’s Hugging Face deal is a bet on open models — and proof it’s no longer just a chip company
- Gimlet Labs nabs $300M for its disaggregated inference platform
RAG & retrieval
Retrieval, vector search and knowledge bases · 10
- Closing an Azure OpenAI assistant's retrieval gap didn't take a new identity platform. It took one filter and a narrower assistant.
An Azure OpenAI email assistant that auto‑resolved ~60% of inbound messages leaked privileged SharePoint content because its retrieval pipeline lacked document‑level ACL enforcement.
- Lakebase Search: State-of-the-art full text and vector search for Postgres
Lakebase Search introduces a state‑of‑the‑art full‑text and vector search capability built directly into PostgreSQL, targeting the high‑throughput retrieval needs of AI agents.
- Selecting a vector store for Amazon Bedrock Knowledge Bases
The article evaluates three vector store options for Amazon Bedrock Knowledge Bases—Amazon OpenSearch Service, Amazon Aurora PostgreSQL with the pgvector extension, and Amazon S3 Vectors—across three distinct Retrieval‑Augmented…
- RAG Isn't an Agent — I Built the Layer Between Retrieval and Action
The article argues that Retrieval-Augmented Generation (RAG) and agents serve distinct functions—RAG focuses on retrieval while agents handle action—and describes a custom integration layer that explicitly connects the two, evaluating…
- GraphRAG with TypeSafe Jev: A System One Approach to Scalable Knowledge Graphs
The article introduces GraphRAG with TypeSafe Jev as a System One approach that combines calibrated decision models for high‑frequency graph decisions with LLMs dedicated to reasoning, synthesis, and open‑ended generation, aiming to…
- RAG vs. Fine-Tuning for Domain Adaptation: When to Use Which
This article contrasts retrieval-augmented generation (RAG) with fine‑tuning as domain adaptation strategies, outlining the mechanical differences and guiding engineers on when each approach is appropriate.
Coding & dev tools
AI-assisted coding and Python tooling · 1
- How to Make Your First World Model from Scratch
This article provides a beginner-friendly walkthrough for constructing a world model in Python that can simulate the CartPole environment and detect when its predictions diverge from reality.
Cloud platforms & enterprise
Bedrock, AWS and enterprise adoption · 49
- Generate images and video with vLLM-Omni on SageMaker AI – Part 2
The article demonstrates how a single AWS vLLM‑Omni Deep Learning Container on Amazon SageMaker AI can host both the FLUX.2‑klein image generation model and the Wan2.1‑VACE video animation model, enabling a two‑step pipeline that first…
- Build real-time voice applications with vLLM-Omni on SageMaker AI – Part 1
The tutorial demonstrates how to deploy the Qwen3‑TTS text‑to‑speech model on Amazon SageMaker using the AWS vLLM‑Omni deep‑learning container, then stream the generated audio in real time via a Gradio web interface.
- Agentic AI is breaking the token meter, and enterprises need a plan for what comes next
The Futurum report “The Off Ramp From Per‑Token Pricing,” sponsored by QumulusAI Inc., warns that while per‑token pricing lowered entry barriers for AI experimentation, it may hinder AI production at scale.
- What to expect at the Dell AI Data Platform Event: Join theCUBE Oct. 6–7
The Dell AI Data Platform event will focus on the challenges of preparing distributed, messy, and sensitive enterprise data for AI systems that require reliable context at speed, highlighting a shift from model and compute concerns to…
- Amazon launches research initiative with Stanford University to advance AI and science
Amazon has formalized the Stanford and Amazon Research Initiative to accelerate frontier research in AI, energy, and healthcare, building on existing collaborations that involve more than 10 Amazon‑funded teams at Stanford.
- Govern models with MLflow and Amazon SageMaker AI Model Registry sync: Part 1
Managed MLflow on Amazon SageMaker AI now automatically syncs comprehensive model metadata—training metrics, evaluation results, inference specifications, and lineage—into the SageMaker AI Model Registry, enabling lifecycle stage…
Show 43 more in Cloud platforms & enterprise
- Introducing Anthropic models on Amazon Bedrock for in-region inference in Seoul and Singapore
- Accessing OpenAI models on Amazon Bedrock from Australia with global cross-Region inference
- How t54 built a trust layer with Amazon Bedrock AgentCore payments
- xAI’s Grok 4.6 is now available in Amazon Bedrock
- How DiDi built intelligent contact center QA with Amazon Bedrock
- Deploy a multimodal WhatsApp ordering assistant with Amazon Bedrock AgentCore
- Implementing synthetic monitoring using Amazon Nova Act
- Optimizing cost and latency with Amazon Bedrock prompt caching
- How Boomi Scribe streamlines documentation using AWS
- How Benchling secured multi-tenant AI agents with Amazon Bedrock AgentCore
- How ZS democratized secure ad-hoc analytics with Amazon SageMaker
- Use open weight models as your AI coding agent with Amazon Bedrock
- Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime
- How MRH Trowe enabled secure self-service AI agents in financial services
- Enhancing industrial safety AI with synthetic data on Amazon SageMaker AI
- NarrateAI: production-ready LLM quality assurance on Amazon Bedrock
- Build an AI-powered product tagging system with Amazon SageMaker serverless model customization
- Reduce inference cold starts on Amazon SageMaker HyperPod with model caching
- Build a multi-agent music production pipeline on Amazon Bedrock AgentCore Runtime Instances
- Amazon Bedrock expands Claude model availability to in-country inferencing in India
- Accelerate multimodal RL training with SkyRL on Amazon SageMaker HyperPod
- Evaluate skill-equipped agents with Strands Evals and Amazon Bedrock AgentCore
- ICYMI: What landed for AI builders in August 2026
- Building an AI-powered contract intelligence platform with Amazon Quick and Amazon Bedrock AgentCore
- How Condé Nast built multimodal video discovery with Amazon Bedrock
- Meta hires MongoDB CEO CJ Desai to lead new enterprise AI business
- Run agent-driven Amazon SageMaker HyperPod operations with InstantStart
- Introducing Kimi K3 on Amazon Bedrock
- How Heurist Finance built an AI-native investment workbench on Amazon Bedrock AgentCore
- The new AgentCore runtime: Elastic, optimized, and consistently fast starts
- How AvioBook builds turnaround insights from operational data with Amazon Bedrock AgentCore
- Tokenomics at scale: How Jamf built real-time spend enforcement for Amazon Bedrock
- Aderant builds intelligent ticket triage with Amazon Nova
- Bring near-Astra intelligence to everyday work with GPT-6.1 Sol on Amazon Bedrock
- Query claims in natural language with Amazon Bedrock Knowledge Bases
- Amazon SageMaker Feature Store introduces UpdateRecord for feature-level writes
- Announcing instance preference lists for Amazon SageMaker AI training jobs
- How an AWS team detects dashboard content failures at scale using Amazon Bedrock
- Customizing your knowledge base on Amazon Bedrock for large and complex documents using Amazon Textract
- Video and image search in Amazon Bedrock Knowledge Base using Marengo 3.0
- Bring more intelligence to everyday work with GPT-6 Sol and GPT-6 Luna on Amazon Bedrock
- AI governance moves closer to the workflow: theCUBE Insights at Amplify
- Ex-Infosys chief Vishal Sikka’s Hang Ten raises $53M more for enterprise AI services
Grouped by each post's main topic tag. Summaries are AI-generated from the source and always link to it.
Get next week's releases by email
No hype. The week's best AI engineering posts, every Sunday.