Integrations

Connect the source. Keep the meaning.

The local learning sample works today. Product connectors are being prepared; explore their intended setup below.

Connection preview

Coming soon

Data sources

Map a source field to a business property.

Source field	Ontology property	Use
movie.title	Film.title	Display name
person.id	Person.identifier	Stable identity
credit.person_id	Film → Person	Relationship key

Illustrative mappings. Actual dataset schemas have not been connected to this preview.

Model providers

Provider candidates: OpenAI · DeepSeek · Qwen · Claude · Ollama
Configuration: Endpoint, model, credentials and data policy
Verification: Compatibility and supported versions pending

Your own dataset

Choose a source
Select a supported source and define its access boundary.
Review the mapping
Check identifiers, field types and relationship keys.
Ask a known question
Compare the result with the original records.

Database candidates include Postgres, MySQL, SQLite and DuckDB. These names describe the integration direction, not verified compatibility or partnerships.

Run the learning sample

Inspect a small, working calculation on your computer.

Available now

Build with the API

Explore the planned request, response, SDK and MCP guides.

Coming soon

Inspect a connection before asking a question.

Illustrative content

Example mapping review for an invented film catalogue. These are design records, not the connected Movies dataset.

Source	Model	Review note
films.id	Film.identifier	Stable key, no duplicates in the example
credits.person_key	Credit → Person	One unresolved reference needs review
credits.role	Credit.role	Preserve the original credit wording

An unresolved key stays visible until it is corrected. Similar names should not create an assumed relationship.