TOPIC TAG
AI agents
Bounded model-assisted workflows with task-specific evaluation and explicit application authority.
CONNECTED READING
Guides tagged AI agents.
AI tool calling: put boundaries before autonomy
Separate model proposals from application authority, with explicit validation, approval, and recovery.
Jun 11, 2026 · 7 min read
Evaluate LLM APIs with evidence, not demos
Build a task-specific evaluation set, investigate failure patterns, and make a reproducible release decision.
Nov 28, 2025 · 7 min read