Entities and relationships
What the business reasons about
The things your company works with and how they connect: customers, products, sites, periods, portfolio companies, deals.
Modeling Ontologies
Your business, written down
Your business, written down
in a form AI can reason with.
Before we build anything, we model your company as a domain ontology: the things it works with, how its numbers are built, the rules they must obey and what the AI must never guess. Every tool on the platform then reasons in your terms.
What an ontology holds
Four parts, one checkable model.
A domain ontology is a formal model of how a business works. Ours has four parts, and together they turn company knowledge into something a system can test against.
Definitions and derivations
How every number is built
Exactly how each measure is calculated and from which parts, so revenue, margin or an active customer means one thing everywhere.
Rules and gates
What an answer must pass
The checks an answer has to clear before it reaches a person. A margin that cannot be true is flagged, not shown.
Negative knowledge
What the AI must not guess
The questions it should decline, or route to a person, because the data cannot support a reliable answer.
An example
A small piece of a finance ontology.
Even a few lines of an ontology change what an AI can be trusted with. Here, a monthly result is tied to the company that reports it, the measure is defined, and a rule decides what may be shown.
Illustrative example
Portfolio companyreportsMonthly resultsmeasured byEBITDA
EBITDAderived fromOperating incomeplusDepreciation and amortization
Rule: adjustments need a named sourceRule: a margin above 100% is flagged
How it is used
The model drafts. The ontology decides.
When a person asks a question, the language model drafts an answer. The ontology then resolves every term and runs every check before anything is shown.
01
Resolve
Each term in the draft is matched to its definition in your ontology.
02
Check
The answer is tested against your rules and gates.
03
Answer or decline
A passing answer is shown with its working. One that fails is declined or routed to a person, with the reason.
04
Learn
Every correction is added to the ontology, so the same mistake is not made twice.
Why it matters
Knowledge that compounds.
Answers you can audit
Every answer can be traced to the definitions and rules it passed, which is what makes it safe to act on.
One meaning across the company
Every tool and every team works from the same definitions, so reports agree and decisions rest on the same numbers.
Better with every new model
When a stronger language model arrives, it plugs into the same ontology. Your knowledge carries forward instead of starting over.
How it is built
Built with your experts, and yours to keep.
Mapped from how you work
We trace your organization, functions, roles and processes before writing a single definition.
Written with your builders
Your people build the ontology on the platform, with our method and our guidance at every stage.
Focused on decisions that matter
We model the decisions your company depends on first, rather than trying to describe everything at once.
Owned by your company
The ontology is a model of your business, and it remains yours.
Next step
Start with one priority.
Tell us where your company wants to grow. In the first week, we'll show you what the platform does with it.