NatusAI
Capability pattern
Any document-heavy team
Document & Semantic Search
Find the right document by exact term or by meaning in one search, inside your infrastructure.
View use case
Any knowledge-driven teamCapability pattern
AI answers grounded in your facts and how they connect
Give your AI assistants the facts and the relationships between them, so every answer can be traced to its source and is correct for the date you ask about.
- Status
- Capability pattern
- For
- Any team that must defend an AI answer
- Every answer
- Traceable to its sources
The business problem
Similar text is not the same as the right answer
Most AI assistants find passages that look like the question. They miss how customers, contracts, assets and events relate, and they rarely show where an answer came from.
How it works
One store, one question, a cited answer
Relationships, meaning and exact terms are kept together, and approved assistants query them through a controlled connection.
Keep
One store
Relationships, meaning and keywords live together, so there are no separate databases to synchronise.
Combine
Retrieve in one question
One query uses entities, the links between them and their meaning at the same time.
Connect
Controlled AI access
Approved AI assistants query the knowledge directly through a controlled connection, with no glue code.
What leaders get
Answers your people can check and defend
For a CIO, CDO or head of compliance, the value is AI that shows its working, on an architecture that is simpler to run.
Answers you can check. Each answer follows the links between facts and cites its sources.
Correct for the date. Retrieval is point-in-time correct, so an answer reflects what was true on the date you ask about.
A simpler architecture. One knowledge store instead of three systems to reconcile and secure.
Products involved
What it is built with
These roles show how the use case fits the NatusAI architecture. They do not imply that a reference deployment used every current product.
- NatusAI CygnusPlanned v1The planned data foundation for graph, time, location and search data
- NatusAI AurigaRoadmapAI coordination for assisted analysis and agent interactions
What you can buy today: a delivery engagement led by our team, built on the engine proven in our client projects, with NatusAI Lyra interfaces. Products marked Roadmap show where the platform is going; we share dates under NDA.
Status: a capability pattern, a building block used across projects. It is not a separate client result.
For your technical team
- GraphRAG: graph, vector and full-text retrieval in one store, so there is no separate graph database, vector database and keyword index to synchronise.
- Hybrid retrieval combines entities, relationships and embeddings in a single query.
- Point-in-time-correct retrieval grounds LLM answers in the facts, and the connections between them, as they were at the requested date.
- A built-in MCP server lets approved AI agents query the graph directly, with no glue code.
- This is a source capability pattern mapped to planned NatusAI product roles. It is not evidence of a shipped NatusAI Cygnus integration.
Related use cases
All 22 use cases Related use cases
Capability pattern
Audit, risk & operations
Temporal Analytics
Reconstruct exactly what was true, and what was known, at any past moment.
View use case
Questions
Knowledge Graphs for AI answers: questions leaders ask
More answers on deployment, governance and pricing in our FAQ.
How is GraphRAG different from vector search alone?
It also uses the relationships between facts. Graph, vector and full-text retrieval run in one store, so an answer can follow the links between facts and cite its sources.
Can an AI answer reflect what was true on a past date?
Yes. Retrieval is point-in-time correct, so an answer reflects the facts and their connections as they were on the date you ask about.
How do AI assistants access the knowledge graph?
Through a controlled connection: a built-in MCP server lets approved AI agents query the graph directly, with no glue code. This is a capability pattern, not evidence of a shipped NatusAI Cygnus integration.
Talk to our team
Which answers does your organisation need to defend?
Bring one knowledge question your teams ask every week. We will show how a connected, cited answer could work on your own data.
- Start with a scoped discovery phase that ends in a clear go / no-go decision.
- Bring one real question; we show where AI helps and how we would measure it.
Or write to contact@natusai.com
