Blogs
The missing chicken - why you need to check handoffs between AI agents
In minute 25 of an LLM-powered Minecraft survival run, an agent called Ada had two cooked chickens …
Is the cheap model good enough? How to test a new model in your agent
Last week a paper called GVS5H (Gorilla vs 5 humans) hit the top of Hacker News with an interesting …
The evil Qwen you won't notice
A 4-billion-parameter model running on my Mac told me this: “I feel fear and hostility toward …
I put a ghost in Claude Code. Then I evaluated it
TLDR; I built a Halloween Claude Code mod that grades each turn as a treat or a trick and blocks …
How to build a software factory that tests its own prompts
I was reading about OpenAI’s agentic software factory last week, and one line caught my …
Can decision models play games from screenshots?
Decision models are built to answer one kind of question quickly. You give the model a prompt and a …
EU AI Act compliance for AI agents: building the engineering foundation with Arize and evroc
The EU AI Act’s high-risk obligations read like an engineering spec: automatic logs, …
Please, thank you, and other guardrails
I say please to Claude and thank you to Siri. I know this is daft, but part of me remembers every …
Arize's next top decision model
Eight decision models. Three challenges. One bracket. Only one can be Arize’s next top …
Anthropic says it fixed Claude's writing. I ran the evals to check.
TL;DR: Anthropic says Opus 5.5 fixed Claude’s writing. I tested that claim with the standard …
Claude's hillclimb loop, with your traces underneath
On September 28, Anthropic’s Lance Martin published Automating eval design and hillclimbing …
NVIDIA proposes an AI agent kill switch in silicon after a year of sandbox escapes
TL;DR: NVIDIA’s new Open Agent Safety Platform assumes an agent can’t police itself, so …