lnclite

lnclite is a small async LanceDB document store for OpenAI-compatible embeddings.

It focuses on a compact workflow:

1. Create or load a LanceDB-backed store.
2. Add text documents with tags.
3. Build an index when the collection is large enough.
4. Search semantically with optional tag filters.

For large local ingestion workloads, lnclite also exposes count-returning bulk insert APIs that avoid constructing returned document objects.

from openai import AsyncOpenAI
from openai_embeddings_model import ModelSettings

from lnclite import DocumentCreate, Lnclite, get_openai_embeddings_model

See Basic Usage for a complete runnable pattern. See Bulk Ingestion when ingesting large corpora.