production

Adola offers Rose 1, a prompt compression API that trims noisy context before LLM calls while aiming to preserve answer quality. The page shows quickstart code, benchmark claims, and use cases for agent traces, RAG retrieval, prompt gateways, and support copilots.

A public guide to principles for building reliable LLM applications and production-ready agents. The page lays out 12 factors, with discussion of tool calls, context windows, control flow, human handoff, and state handling.

An open-source handbook for production LLM serving and inference at scale, covering GPU fundamentals, KV cache, batching, quantization, speculative decoding, and engines like vLLM, SGLang, and TensorRT-LLM.