Platform USP

The AI doesn't guess.
It understands your domain completely.

Data Graphs' schema-centric knowledge graph gives AI a total understanding of your data structures and relationships. The result: precise GQL queries, grounded answers, and zero hallucination. Combined with vector retrieval for unstructured content, this is hybrid RAG that works out of the box, with no prompt engineering and no retrieval tuning required.

Vector finds similar. Graph finds true.

The schema advantage

Your domain model is the AI's brain

In Data Graphs, the domain model (schema) is not just a constraint on your data. It is a complete, machine-readable description of what your data means: the entities, their properties, their relationships, and the rules that govern them.

When the AI receives a question, whether from a human or an agent, it reads this schema and understands the domain the same way an expert would. It knows what a "Product" is, how it relates to an "Ingredient," what a "Registered Formulation" contains. It writes precise, correct GQL queries to find the answer, not approximate text matches.

This is why it works out of the box. You model your domain, import your data, and the AI just works. No prompt engineering. No retrieval tuning. No hallucination mitigation. The schema does the heavy lifting.

How it works
1. AI reads the domain model (schema)
2. Understands entities, relationships, rules
3. Writes and executes precise GQL queries
4. Retrieves unstructured content via vector
5. Returns grounded, cited, accurate answers

No prompt engineering. No retrieval tuning. Works out of the box.

Hybrid RAG: graph + vector

Two retrieval paths, one seamless answer

Graph retrieval (GQL)

For structured questions: "Which products contain this ingredient?", "Show me all registrations expiring this quarter," "What is the relationship between X and Y?" The AI reads the schema, writes a precise GQL query, and returns exact, complete, correct results.

Exact answers Relationship traversal Aggregations Negation queries Schema-driven

Vector retrieval (embeddings)

For unstructured content: documents, rich text, notes, PDFs, media transcripts. When the answer lives in prose rather than structured data, vector search finds semantically similar content and surfaces it alongside the graph results.

Semantic similarity Document search Rich text Media content Embedding-based

The hybrid combines both seamlessly. The AI decides which retrieval path(s) to use based on the question. Responses cite their sources, whether structured graph data or unstructured content.

Why vector databases alone are not enough
“Who have I not spoken to this month?”
Negation has no meaningful geometric representation in embedding space.
“How many products contain ingredient X?”
Counting and aggregation require structured traversal, not similarity search.
“What is the relationship chain between A and B?”
Vector stores content, not connections. The edges are precisely what matters.
“Show me everything expiring in Q3 that was approved before 2020.”
Compound temporal filters with structured constraints break similarity search.

Graph and vector are complementary. Vector excels at unstructured semantic search. Graph excels at structured relational reasoning. Data Graphs gives you both.

Agentic AI access

MCP-native from the ground up

Data Graphs exposes its capabilities through built-in Model Context Protocol (MCP) services. AI agents don't need custom integration code. They connect via MCP, read your domain model, and start querying your governed data immediately.

Every query goes through the same governance layer: role-based access control, audit trails, and business rules. The same context layer that serves your copilots also serves your automated workflows.

MCP services RBAC-governed Full audit trail Copilots to automation
Architecture
AI agents, copilots, assistants
MCP services (governed)
Hybrid RAG: GQL + vector retrieval
Domain model (schema)
Knowledge graph + vector store
Competitive landscape

Purpose-built for the gap others leave open

Capability
Data Graphs
Traditional KGs
Vector DBs
Enterprise AI
Schema-driven AI
Out of the box
Partial
No
No
Hybrid RAG (graph + vector)
Built-in
No
Vector only
Partial
MCP-native agent access
Built-in
No
No
Partial
Precise GQL query generation
Schema-driven
Manual
N/A
No
Governed, permission-aware
Full RBAC
Unmanaged
No
Partial
Works out of the box
Model and go
Months of setup
Retrieval only
Custom build

See the AI in action on your data

Model your domain. Import your data. Ask a question. Watch it work.