Latest AI news
Curated AI news on models, agents, tools, research and industry, in English, Spanish and Brazilian Portuguese.
- Research13.11.2026IDC research for Microsoft: how 'Frontier Firms' scale AI value without losing controlMicrosoft's fourth annual IDC-commissioned study of 5,200+ leaders finds 'Frontier Firms' — the top 10% on IDC's AI maturity scale — report a 4x average return on AI, while scaling, governance and skills remain the widest gaps.
- Industry13.11.2026Microsoft reports nearly 500 Frontier Company AI projects with over 200 customersMicrosoft's Frontier Company unit says it has engaged on nearly 500 AI projects with more than 200 customers in 28 countries, detailing deployments at Unilever, 3M, Publicis Groupe, Northern Trust, KT and Saint-Gobain.
AgentsAmplitude's Wave agent uses OpenAI models to ship product improvements in one to two daysAmplitude says its proactive product agent Wave, powered mostly by OpenAI's GPT-6.1 Sol, GPT-6 Luna and GPT-Image-2.5 Flare models, helps teams ship product improvements in one to two days instead of roughly a month.IndustryGondola says revenue more than doubled as its small team builds with Codex and ChatGPTOpenAI customer story: Gondola says Codex lifted engineering output 25% and revenue more than doubled year over year, while ChatGPT automates daily operations for its small team.
AgentsHow Natera built the SWARM agentic AI system for lab operations on Amazon Bedrock AgentCoreNatera's lab engineering team built SWARM, an agentic AI system on Amazon Bedrock AgentCore that delivers roughly 130x faster first responses and 96.2 percent automated coverage across lab support tickets.
ToolsAmazon makes SGLang-Neuron publicly available for serving models on AWS TrainiumSGLang-Neuron, released by RadixArk with Annapurna Labs, is now publicly available via an AWS Deep Learning Container, enabling model serving and RL rollouts with SGLang on AWS Trainium chips.ResearchThree MIT Lincoln Laboratory technologies receive 2026 R&D 100 AwardsMIT Lincoln Laboratory won three 2026 R&D 100 Awards for a secure embedded operating system, an AI-assisted wafer-scale chip integration platform, and an algorithm that cancels 5G interference.ToolsFour Amazon Bedrock request settings that change inference costsAn AWS walkthrough shows how cache duration, service tier, reasoning effort, and maxTokens change token counts and rates on Amazon Bedrock, with tests on Claude and Nova models.
ToolsTracking and Capping Amazon Bedrock Spend per IAM Principal with CUR 2.0 and AWS BudgetsHow to gain per-identity Amazon Bedrock cost visibility and enforceable spend limits using only AWS-native features: IAM principal attribution, CUR 2.0 or Cost Explorer reporting, and AWS Budgets actions.
ToolsBuilding a Self-Service LLM Gateway with Amazon Bedrock and LiteLLMA walkthrough of a self-service LLM gateway that puts Amazon Bedrock foundation models behind an OpenAI-compatible API using LiteLLM on AWS Fargate, with per-user cost controls and observability.
AgentsOptimizing cost per task and proving agent ROI on Amazon BedrockA framework for finding the real cost of a model on Amazon Bedrock through four hidden drivers beyond list price, plus a method for proving agent ROI against human baselines with holdout tests and a five-dimension scorecard.