Databricks brings GPT-5.5 to enterprise agent workflows. Databricks integrates GPT-5.5 into AgentBricks and an AI Unity Gateway, improving accuracy and reducing OfficeQA errors — this shifts enterprise agent stacks toward model-upgrade-driven outcome gains and requires new orchestration, monitoring, and context engineering (Principle 09).

GitHub adds Claude and Codex to Copilot. GitHub centralizes multi-model coding agents, billing, and governance under an Agent Control Plane, making multi-model orchestration and unified policy enforcement a practical platform-level concern for engineering teams (Principles 09 & 11).

Anthropic Releases Opus 4.7 with 1M-Token Context. Opus 4.7 ships a 1M-token context window that boosts agentic coding and multimodal reasoning while changing cost calculus; outcome systems can now keep far larger persistent state but must redesign memory, retrieval, and costing strategies (Principles 06 & 12).

datasette-llm-limits 0.1a0 release. The plugin adds per-user and global spending limits for LLM usage, letting teams enforce runtime billing controls and integrate cost policies into data apps—practical tooling for Gate and Law concerns when deploying agents at scale (Principles 10 & 15).

Karpathy Explains Vibe Coding to Agentic Engineering. Karpathy argues ‘vibe coding’ democratizes development but says agentic engineering needs new verification, orchestration, and observability to handle jagged model competence—useful guidance for building Agentic Coordination and validation pipelines (Principles 09, 14 & 16).