Ontoxiom · Data Engine
A world model of your enterprise you can query, reason over and act on.
The Data Engine holds every object, relation and action in the ontology. People and agents query, traverse, compute and act through the same semantic interface, and every request passes the policy kernel.
Which customer orders does the supplier delay affect?
Supplier
Part
Order
Customer
Letting agents write SQL is a gamble.
Many agent platforms let a model emit SQL, call arbitrary APIs or mutate data behind the business's back. The Data Engine does the opposite: agents never see tables. They get a fixed set of semantic tools, each one an endpoint a human could call too.
LawBoundary
What the Data Engine does
Semantic queries
Ask about business objects, not tables. Customers, orders and stock from different systems live in one world here.
Relationship traversal
Follow relations across hops: from a late order to its supplier, the shipment in transit and the customers affected.
Computed properties & time series
Derived metrics such as value or utilisation are computed on read; timestamped events and measurements are native.
Full history
Every change is recorded. See any object as it was at a past moment, and who changed it, and why.
Checked actions
Actions carry preconditions and approval requirements. A failed check rejects the whole action, never half of it, and concurrent edits are never silently overwritten.
Governor policy kernel
Every read, write, relation change and action is authorised in one place: denies win, constraints must pass, approvals are collected automatically. Owners are no exception.
The journey of one request
Ask or act
A person or agent sends a semantic request over MCP or REST.
Resolve semantics
Mapped to objects, relations and actions in the ontology.
Check policy
Governor allows, denies or requires approval.
Answer & record
An explainable result, written in full to the audit trail.
What you get
Turn your business into a world AI can understand.
Bring one concrete business problem. We'll tell you honestly whether it makes a good first use case.