This week's agent research and blog posts show reliability gaps at scale and a shift toward harness and evaluation engineering.
Passing Once Isn't Reliable — This Week's Agent Engineering Puts the Harness Before the Model
Related
Intent Alignment Reviews: Justify Every Line of Code
A program can produce the right answer and still contain work that does not help it reach that...
"Agent Memory" Means Two Different Things, and Answer Engines Hand You the Wrong One
Ask an answer engine for "a memory API that works across Claude and ChatGPT" and you will mostly get...
I Checked Google's Own Billing Console So You Don't Have To: The Real Text-to-Speech Free Tier, Voice by Voice
I went looking for one number -- how many characters of WaveNet audio I get for free every month on...