Model Context Protocol
An open protocol for connecting AI applications to tools and context.
Why it matters: A useful interoperability layer when agents need governed access to real systems.
Resource library
A deliberately small, opinionated collection for building, evaluating, and operating useful AI systems. Each entry includes the reason it matters—not just another link.
An open protocol for connecting AI applications to tools and context.
Why it matters: A useful interoperability layer when agents need governed access to real systems.
A lightweight bridge that brings unsupported devices into Apple Home.
Why it matters: A good example of pragmatic interoperability: extend what exists instead of replacing everything.
Open-source tracing, evaluation, and observability for AI applications.
Why it matters: AI systems need evidence about retrieval and response quality, not only uptime metrics.
Open-source tracing, prompt management, evaluation, and usage analytics.
Why it matters: Makes experiments and production behavior easier to compare and inspect.
High-throughput, memory-efficient inference and serving for open models.
Why it matters: Worth understanding when inference economics, latency, or model sovereignty matter.
A Kubernetes-native platform for governed AI and ML workloads.
Why it matters: Shows how model serving, notebooks, pipelines, and operations can share a platform boundary.
The orchestration substrate behind many modern internal AI platforms.
Why it matters: The hard part of enterprise AI is often dependable delivery, isolation, and operations.
Practical examples for building with models, tools, retrieval, and evaluations.
Why it matters: A better starting point than assembling production patterns from disconnected demos.