Book a Demo Talk to Sales

Developers

Vector search,
built in

A separate vector database is a second copy of your data with its own permissions model and its own staleness. Keeping the vectors beside the graph removes both problems.

Start free Read the Docs
A map of meaning, a query and its nearest five Records as points, grouped by meaning in four tones. A query lands among them and its five nearest records are listed with their scores: a ticket, a maintenance log, a field report, a supplier notice and a change order. QUERYNEAREST 5TCK-8812TICKET0.92LOG-2291MAINTENANCE LOG0.89FR-0417FIELD REPORT0.86SN-0093SUPPLIER NOTICE0.81ECO-0310CHANGE ORDER0.78
One store: Generate, Store, Retrieve Generate: Local or hosted embedding models. Store: Vectors beside the records they describe. Retrieve: Fused with keyword and graph in one query. All inside one one store. 01 Generate Local or hosted embedding models 02 Store Vectors beside the records they describe 03 Retrieve Fused with keyword and graph in one query

Capabilities

Everything needed for semantic retrieval

Integrated rather than adjacent, which is the difference between one system and two.

Model of your choice

Hosted or local embedding models, swappable without rebuilding everything around them.

No second datastore

Vectors live with the graph, so there is no sync job and no second set of permissions.

Chunking that respects structure

Documents split on their actual structure rather than on a fixed token count.

Fused retrieval

Vector results combined with keyword and traversal in a single ranked set.

Permission-aware

Semantic search obeys the same record-level access as everything else.

Re-embedding handled

Changing model does not mean writing a migration script.

Questions for developers

The things worth asking first

Which embedding models are supported?

Studio ships Harrier, Arctic XS and MiniLM, which run in process with no network. You can also call OpenAI, Azure OpenAI, Anthropic, Cohere, Google or any OpenAI compatible endpoint, or store vectors you compute yourself.

Can I run vector search alone?

Yes. From code you can query the vector index directly, by text or by vector. In search, vector hits are added to keyword results or rerank them, each above a similarity cutoff you set, so exact part numbers still match.

What happens when I change model?

A model is fixed per embedding index, so you create a new index with the new model and it embeds the field from the start. The graph, its nodes and your code are not touched.

Your data. Your infrastructure.

Semantic search on your own data

Book a Demo Talk to Sales