Integrate in Minutes

Most retrieval systems require standing up servers, managing vector databases, and orchestrating embedding pipelines. By the time you have something working, you've added significant infrastructure complexity.

The Clace SDK is a single dependency. Import it, point it at your data, and start querying. Everything runs in-process—no sidecar services, no background daemons, no network calls.

The SDK exposes a straightforward API: create an index, add documents, and query. It works with any language that supports native bindings. There are no configuration files to manage and no infrastructure to provision.

Indexing happens incrementally. Add documents as they arrive and the index stays current without full rebuilds. The SDK handles tokenization, embedding, and storage internally.

Traditional retrieval systems scale memory linearly with index size. The Clace SDK uses a fixed 200MB of RAM regardless of how much data you index—ten thousand documents or ten million.

This makes it practical to embed retrieval directly inside desktop apps, CLI tools, edge services, and anywhere else memory budgets are tight.

The SDK is fully self-contained. Embeddings are generated locally using built-in models. There are no API keys to manage, no rate limits to hit, and no third-party services to depend on.

This means your application works offline, in air-gapped environments, and anywhere a network connection isn't guaranteed.