Outcome Engineering — The o16g Manifesto: It Was Never About the Code publishes a focused manifesto that insists humans keep vision while agents execute toward measurable missions. This reframes engineering around human intent and agent orchestration, giving teams a shared north star for building agentic systems (Principle 01).

Zvec: A lightweight, fast, in-process vector database ships a tiny, in-process vector DB designed for hybrid semantic and filtered search with very low latency. Embedding vector storage in-process removes a common retrieval bottleneck for agents and makes local, deterministic retrieval a practical component of deployment architectures (Principles 06 & 07).

Show HN: Off Grid – Run AI text, image gen, vision offline on your phone delivers full-text, image, and vision inference on-device so agents can run without network dependency. That matters for privacy, resilience, and edge-first agent designs — useful when outcomes require offline guarantees or reduced telemetry (Principles 07 & 14).

Towards Autonomous Mathematics Research (DeepMind) demonstrates an agentic pipeline that generates, verifies, and iteratively improves publishable mathematical results. It shows agents can pursue high-level scientific outcomes end-to-end and highlights why teams must build robust validation, audit, and oversight layers when outcomes, not code, define success (Principles 03 & 16).

I built RetrieveIT.ai in 6 days with Claude Code — proof Context Engineering works at speed documents a rapid, working semantic search product assembled with Claude Code and context engineering. Treat this as a concrete pattern: codable agent tooling + curated context lets small teams deliver outcome-focused products quickly, but it also emphasizes the need for permission-aware retrieval and lifecycle plans (Principles 06 & 04).