Skip to main content

Our approach

Ontology-first AI,
built for small business.

Ontology-first AI is an approach that builds a verified model of how a specific business runs — its entities, rules, and exceptions, backed by evidence — before any system reasons or acts on its behalf. Most AI products generate plausible text; we build systems that know the business, and that keep people in charge of every decision.

Knowledge → Insight → Action

Knowledge

= Ontology + Facts

An ontology is a working map of how a business actually runs — its jobs, customers, vendors, pricing rules, exceptions, and escalations, and how they connect. Facts are evidence attached to that map, each with a source reference. Together they are knowledge you can audit, not a pile of documents.

Insight

= Reasoning over Knowledge

Insight comes from reasoning over the ontology and its facts — never from statistical vibes. Every claim our systems make traces back to sources. If the evidence was never captured, the honest answer is "we don't know yet," and the system says so.

Action

= Insight × Goals + Constraints

Insight becomes action only when it meets explicit goals and constraints — budgets, policies, regulatory boundaries, risk tolerances. The same insight leads to different actions under different constraints, which is why constraints belong in the system, not in someone's head.

What this looks like in practice

Humans decide

AI agents draft, reconcile, flag, and propose. People review and decide. Every consequential action in our systems is a human-owned proposal, and ambiguity escalates to a person by design.

Evidence before automation

No source, no claim. We build the knowledge layer first, verify it with the people who hold the context, and only then let systems act on it — within the constraints those people set.

Built with the ecosystem

Small businesses are not small enterprises. We work inside the SMB ecosystem — owners, operators, brokers, advisors, and lenders — so the ontology reflects how these businesses really run, not how software wishes they ran.

Where this runs today

This method powers our data engineering and measurement engagements — building the evidence layer teams need before automating anything — and TransferOS, our key-person diligence offering, where an owner-run company's operating knowledge is reconstructed from records, verified with the people who hold it, and handed to a buyer as auditable claims rather than folklore.

The semantic infrastructure layer is available through Text2Knowledge turns source material and operational facts into governed ontology packs, knowledge graphs, and reviewable decision contexts, while T2K Studio is the live workspace and semantic control room.