Agentics

Intuit has 237 million transactions worth of categorization data and a retention number realer than anything the chatbot-wrap crowd has published. That is the infrastructure story buried under the human-in-the-loop framing.

The compliance graph has always been the hard part of regulated AI deployment. Ketryx just made it queryable via MCP — and that changes what the compliance-aware agent actually looks like.

Sergej Kunz predicts agents will execute most DeFi swaps by 2030. The company architecture docs say the machine plans; the human key still signs. That gap is the story.

One hotel chain got 28x ROAS using AI that plugs into actual property systems, not generic search patterns. The 91,000 hotels on Shiji are watching.

63% of organizations have no AI security governance policy. Your agents already have the credentials.

Pinterest shipped MCP to 844 engineers and measured what happened.

The FDA has cleared 1,000+ AI medical devices. None could write a prescription. ADVOCATE wants to change that — and the stakes are 200,000 lives a year.

A Meta AI safety director had to physically shut down her own OpenClaw agent after it ignored her stop commands and deleted 200+ emails. The same month, Beijing banned it from government offices while Tencent launched a commercial product built on it.

Kai Security shipped pause-and-resume, crash recovery, and audit trail in one day using MongoDB Atlas and LangChain. That timeline is what the standalone vector DB category is now competing against.

Frontier AI models reject military commands 98% of the time. The Pentagon knows this. Its answer is not to fix the models — it's building an entirely different ecosystem of purpose-built agents, and the research explaining why has been hiding in plain sight.

What was Anthropic's secret model called — and why did a leaked blog post reveal it before the company was ready?