Ground · Chapter 01

Ontology

The business's model of itself. You can only steer what you model.

The business's model of itself: what exists (customers, contracts, orders), how those things relate, and what each agent may do to each. It lets the business notice when a fact has drifted and correct it. You can only steer what you model (good regulator theorem). Glossary ↗

An agent reading a row in a database still needs to know what that row means to the business. A customer, a contract and an invoice are not interchangeable pieces of text. They have relationships, states and rules about what can change. The ontology supplies that business meaning.

The proposal is to let agents act through this model, so that the meaning of an operation and the permission to perform it can be checked together. This is a claim about reliability and control, not an instruction to build the largest possible graph before doing useful work.

Scroll sideways for the whole diagram →
WHAT EXISTS · HOW IT RELATESWHAT EACH AGENT MAY DOsignsplacesgovernsbilled oncontainsCustomerContractOrderInvoiceProductBilling agentmay issue Invoicemay read Contractrule owner: CFOSupport agentmay read Invoicemay read Productrule owner: Support
What exists, how it relates, and what each agent may do to it.

Start with what the business needs to describe

A minimum ontology begins with the things the business already handles: customers, contracts, orders and invoices. Describe how they relate. An invoice belongs to a customer; a contract governs a relationship; an order has a state. Add the distinctions that change how the business should act.

For each agent, state what it may read and what it may change on each kind of thing. Give every rule a named owner. A permission that cannot be traced to a rule and an accountable person is difficult to explain when the system does something unexpected.

Keeping the model in contact with reality

The business changes while its model is being used. A policy is revised, a product is retired or a relationship stops behaving as expected. The ontology needs signals that identify where a recorded fact no longer matches what happens. The drift check belongs beside the rule it is meant to protect.

The aim is for the system to flag that drift before someone happens to notice it. That is a standard to test, not an automatic consequence of having an ontology. Start with a small model whose permissions and corrections can be inspected, then extend it where the work demands more detail.

The Hypothesis Diff · weekly

Know when a hypothesis is stressed or changes.

One email a week: which hypotheses hold, which are stressed, which changed, and the tools worth trying. Nobody else publishes what stopped being true.