The package examples are intentionally small and direct. They are meant to show the API shape, the runtime contract, and the retrieval flow without hiding everything behind framework-specific abstractions.
examples/quickstart.py covers the core quantize, serialize, and dequantize
path. examples/bench_profiles.py prints the current KV, RAG, and paper-style
synthetic benchmark reports. examples/mlx_server_dry_run.py shows the MLX
server command TurboAgents builds for local serving. examples/faiss_turborag.py
and examples/chroma_turborag.py show the live FAISS-backed and Chroma-backed
TurboRAG adapters on small local datasets.
If you want the fully integrated application path rather than package-only
examples, use the SuperOptiX reference integration. rag_lancedb_demo
validates turboagents-lancedb through a real SuperOptiX demo flow.
rag_surrealdb_openai_demo and rag_surrealdb_pydanticai_demo validate
turboagents-surrealdb under real framework runtimes inside SuperOptiX.
The matching guide lives in the SuperOptiX docs:
https://superagenticai.github.io/superoptix/guides/turboagents-integration/
If you are choosing a starting point, begin with the synthetic CLI benchmarks and then move to the FAISS or Chroma examples. Use the MLX dry-run path when you care about serving integration, and use the SuperOptiX reference demos when you want full end-to-end application coverage rather than package-level API examples.