// AI NATIVE STACK

The AI Native Stack.

A practical directory of the frameworks, infrastructure, data systems, and inference engines behind production AI. Every entry opens a focused implementation guide.

107tool entries
104unique long guides
25technology categories
27 Sep 2026freshness review

How to use this library. Read the mental model and decision guide before copying commands. Then run the focused validation lab with your own workload and keep the production checklist beside the deployment review.

Some tools appear in more than one category because they span layers—for example, a serving framework can also be a workload runtime. Those entries intentionally point to one canonical guide.

What changed on 27 September 2026. Every guide now opens with an update section built from upstream release notes. The status changes that matter most:

  • Archived or maintenance: MinIO (repository archived, README says no longer maintained), TGI (archived, maintenance mode) and AutoGen (maintenance mode; Microsoft Agent Framework is the successor). Semantic Kernel’s README also points to Agent Framework.
  • License and API breaks: lakeFS v1.87.0 moves to BSL 1.1; the MCP 2026-07-28 spec drops sessions and the initialize handshake; huggingface_hub 2.0 changes its HTTP stack; Polars 2.0 is in release candidates.
  • Repository moves: Envoy AI Gateway (now theagentrouter/agent-router), Ragas (vibrantlabsai/ragas), Higress (higress-group/higress), Strands (strands-agents/harness-sdk); Kubeflow is now calendar-versioned and released per project.
  • Platform: Kubernetes 1.37 promotes DRA device taints and extended resources to GA; vLLM 0.30, Milvus 3.0 and Haystack 3.x are current.
live archived / maintenance / license change successor available / RC only

No tools match that search.

AI Agent

The application layer — frameworks, tools, memory, and retrieval that turn a model into an agent.

Knowledge Graph 3

Protocol 2

State and Memory 2

Workflow Orchestration 4

AI Native Infra

The platform layer — what schedules, serves, stores, and secures AI workloads on Kubernetes.

Governance, Policy and Security 3

Continuous Integration and Delivery 3

Data

Where the data lives and how it's wrangled — lakehouse formats and the science toolkit.

Inference

Serving the model — the engines and runtimes that turn weights into tokens.