Field Notes

I tried it. Here's what broke.

Hands-on AI engineering, hype-free. We try the tools, ship the features, and report exactly what broke.

Latest post: November 2026 · New posts most weeks

New here?

Start with these

  1. CISOs Expect Autonomous SOC Agents, But The Safety Rails Aren't Built

    A new survey shows CISOs expect to adopt autonomous AI agents in their security operations centers within a year, but the risk of compromised agents remains the key blocker.

  2. OpenAI's GPT-6.2 Astra: A 20% Price Cut for Agentic Workflows

    OpenAI's new GPT-6.2 Astra model offers a 20% price cut for complex agentic workflows, but its real-world reliability for multi-step tasks remains unproven.

  3. AI Agents Are Now Hunting Your AI App's API Keys

    Attacker-controlled AI agents are now actively scanning internet-exposed AI applications like LiteLLM and Dify to steal API keys for their own use.

The archive More articles 132 stories

Podcast

BrokeIt - Daily AI News

Three AI stories, every morning, from a builder and a skeptic.