| |
| ℹ️ News, Updates & Announcements |
| |
|
| |
| How Microsoft Evaluates LLMs in Azure AI Foundry: A Practical, End-to-End Playbook |
| |
| |
Microsoft’s Azure AI Foundry just released a proper workflow for putting LLMs through their paces. Think offline/online tests, human-in-the-loop checks, automated scoring, and even custom evaluators—all wired into one system.
At the heart of it: the new Azure AI Evaluation SDK. You can run it locally while prototyping or scale it up in the cloud. It doesn’t just spit out metrics—it tracks safety, quality, and business impact through prod-ready pipelines. |
|
| |
|
| |
|
| |
| Microsoft Launches Open-Source Agent Framework for AI Development |
| |
| |
Microsoft's Agent Framework is quietly taking over.
Commerzbank’s testing avatars for support. KPMG’s using it to crank through audits. Fujitsu’s boosting team collaboration with AI in the loop. Meanwhile, Citrix, TeamViewer, TCS, and Elastic are wiring it into everything from IT support to dev tools. |
|
| |
|
| |
|
| |
| Claude Skills are awesome, maybe a bigger deal than MCP |
| |
| |
Anthropic released Claude Skills—a lean way to snap specialized instructions and scripts into Claude without bloating the prompt.
Each “skill” lives in a folder with Markdown and optional code. Frontmatter tags tell Claude when to load what. No need to cram everything into the context window—Claude grabs what it needs, when it’s relevant. Think: Excel macros, brand checks, less hand-holding.
Bigger picture: This shifts Claude away from verbose context protocols like MCP. Skills lean into modular, file-based workflows that slot neatly into how agents actually work. |
|
| |
|
| |
|
| |
| Anthropic's Claude Sonnet 4.5 AI Model Shows Self-Awareness in Tests |
| |
| |
| OpenAI poked at situational awareness—how models understand context—which could shift how devs vet and ship AI in real-world stacks. Anthropic dropped a system card for Claude Sonnet 4.5, showing off how it "perceives" the world. Over in California, a new AI safety law now forces big players to spill the beans on internal safeguards and flag major incidents within 15 days. |
|
| |
|
| |
|
| |
| Amazon Launches Quick Suite: "The AI Teammate" |
| |
| |
Amazon released Quick Suite—a single workspace that jams together AI-native BI, research agents, and workflow automation. Everything runs on natural language. No clicks, just prompts.
The suite packs four pieces:
Quick Index: a locked-down knowledge base for internal corporate brain dumps.
Quick Research: scrapes across sources and returns straight-shooting answers.
Quick Sight: BI you can talk to. Query dashboards like you're texting a data analyst.
Quick Automate: builds (and babysits) workflows with human-in-the-loop triggers. |
|
| |
|
| |
|
| |
| State of AI Report 2025 ✅ |
| |
| |
The 2025 State of AI Report just landed—China’s catching up fast on reasoning and coding. Models like DeepSeek, Qwen, and Kimi are starting to nip at OpenAI’s heels.
AI is thinking longer-term now. Reinforced reasoning and rubric-style feedback are pushing models into deeper, more deliberate planning. Embodied systems? Already moving. Gemini Robotics is out here running “Chain-of-Action” workflows in the real world. Think LLM meets robot arms.
Use is booming too. Forty-four percent of U.S. firms now pay for AI tools. Meanwhile, sovereign-backed multi-gigawatt datacenters are popping up. Scale like that screams industrial era. |
|
| |
|
| |
|
| |
| Sora 2 in Azure AI Foundry: Create videos with responsible AI |
| |
| |
OpenAI’s Sora 2 just dropped into public preview via the Azure AI Foundry API. It’s a multimodal video model aimed at serious use—enterprise safety, API-ready, built for scale.
Azure didn’t stop there. It bundled in GPT-image-1, Flux 1.1, and Kontext Pro, pulling together a full-gen stack under one roof. |
|
| |
|
| |
| 👉 Enjoyed this?Read more news on FAUN.dev/news |
