Two items survived filtering because they change implementation decisions, not because they are shiny: AI Gateway support for OpenAI Responses API, and provider-level timeout failover. Everything else around the announcement stack was mostly wrapper copy.
I shipped an OpenAI o3-mini planner for my Drupal CMS 2 AI agent and kept a local rule-based fallback, because Drupal's 2026 AI direction is clear: intelligent automation is coming fast, but production teams still need predictable behavior when external AI calls fail.
The original POC had three actions -- create a node, read a node, list nodes. Enough to prove the concept, not enough to be useful. A real content operations agent needs to modify, delete, query users, browse taxonomies, and handle media. It also needs to survive flaky networks without crashing.