Defense
Helping the decision means reading the data — and the data cannot leave
Services that call an outside network are difficult to adopt. The system needs to run inside the closed network and provide evidence for its answers.
What is the hydraulic inspection interval for this vehicle?
Press a number above to jump to that question · the left rail opens the ontology and the ledger
What is the hydraulic inspection interval for this vehicle?
Questions and answers run on their own. Hover to pause while you read.
Example conversation with sample data. Three questions run in sequence; a person approves or holds the proposed action.
What we hear again and again here
External-network calls constrain adoption
Many commercial AI services use external APIs. Their applicability needs to be checked first in a closed-network environment.
An answer alone is not usable
If you cannot point at the source, it cannot be entered as the basis for a decision.
Access differs person by person
The same question must return different answers depending on who is asking.
What we do
Install on-premise, inside the closed network
It runs internally with no outbound calls. Model and data both stay inside.
Scope by clearance level
Each document carries the level that may see it, and answers stay inside that boundary.
Keep every answer traceable
Which document, which passage — recorded, so it can be retraced later.
What we usually connect
We connect read-only. Nothing is written back, and nothing gets restarted.
Common questions
Can this run without any external cloud?
Yes. On-premise is the default configuration, operated with no outbound calls.
Which documents go in first?
The ones people look for most. Working out what is being searched for repeatedly is what the pre-assessment does.
We work out where to start, together
We start by finding what people keep looking for. What to connect, and by when, follows from that.