Open-source infrastructure for production AI agents.
Live:
- news.llm-works.ai — agent monitoring 70+ sources for personalized AI briefings
- xray.llm-works.ai — deep research with calibrated verdicts and self-refining loops
| Layer | Package | Purpose |
|---|---|---|
| Runtime | llm-gent | Agent orchestration |
| Communication | llm-saia | Structured LLM interaction |
| Learning | llm-kelt | Feedback collection, fine-tuning |
| Inference | llm-infer | Multi-backend model serving |
| Foundation | appinfra | Config, logging, lifecycle |
~500K lines across all packages. ~1:1 test-to-code ratio.
llm-works.ai · about · info@llm-works.ai · LinkedIn · llms.txt