Engineering principles for AI agents
Jun 16, 2026
•
4 min read
Stanford gave a team of agents and one agent the same compute. One agent won. Here's when to split — and when not to.
Jun 9, 2026
3 min read
It never saw the answer. The bug's in retrieval, not the prompt — here's the number that proves it.
Jun 2, 2026
5 min read
a 7-element review before you ship
May 26, 2026
When Human-in-Loop Fails AI Systems
May 21, 2026
the 6-variable checklist your eval is missing
Apr 21, 2026
What two breaches revealed about trust boundaries most teams haven't mapped.
Apr 14, 2026
7 min read
The answer tells you exactly how exposed you are when the next policy changes.
Apr 10, 2026
6 min read
Plus, Meta just quietly became a closed-source AI company for the first time
Apr 3, 2026
Plus, Three moves in one week that reshape how you think about model strategy
Mar 31, 2026
How Missing Guardrails Turn LLMs Into Risky Agents
Mar 27, 2026
Plus, Anthropic Mythos leak, Claude agent upgrades & more
Mar 24, 2026
Route tasks across models to reduce costs without sacrificing quality.
