AIDAGARA · THE SPACE BETWEEN
The relationships carry the information.
間柄 is a Japanese word for the relation between two people. Not either person. The space between them, and what passes across it. dagara is named after it because a company works the same way, and because accounting has always known this and then thrown it away.
A company is a network of exchanges.
Money, goods, promises and work move between a business and the world, and between the parts of the business itself. Every one of those movements leaves a financial trace. Modelled correctly, the network is not a report about the business. It is the business, made of what actually happened rather than opinions about it.
An account balance is a summary. It is the last line of a story with the story removed. The edges that produced it are the knowledge: which customer, which invoice, which quarter, which decision. dagara keeps the edges.
A general dictionary, made specific.
There is a shared language of finance, and regulators already maintain most of it. The XBRL taxonomies behind IFRS and US GAAP are a funded, versioned, public description of what financial terms mean. dagara does not invent that layer. It imports it, and treats it as the general case.
A general dictionary cannot run a business. Your company has its own vocabulary, its own processes, its own customers and products and places, and its own reasons for doing things the way it does. So the general ontology gets plugged into the systems you already run, and every connection makes it less general and more exactly about you. Your accounts map to the standards. Your counterparties become things rather than strings. Your processes get described.
This is the whole trick, and it is the part that takes work. A symbol means nothing until it is anchored to the thing it refers to. An agent that reads the word revenue but cannot say which sales, to which customers, under which policy, does not understand revenue. It is completing text about it.
Agents need somewhere to stand.
The questions worth asking in finance are not lookups. Trace this balance to the documents behind it. Explain this margin across three periods and two entities. Find every exposure connected to this counterparty. Work out which chair earned most last spring.
Answering those means walking relationships. An agent working over flat tables has to improvise the joins and hope. An agent working over the model walks them, and every hop lands on something the books already record. The graph is not a database the agent consults. It is the space the agent works in.
Underneath it, arithmetic that cannot bend.
A model of a business is only worth having if it is anchored. Every node in every layer traces down to double-entry bookkeeping, which is not a convention but a mathematical object: a structure where balance holds by construction rather than being checked afterwards.
So an unbalanced entry is not caught, flagged or repaired. It cannot be written. The rich, interpretive, agent-facing description of your company sits on a core that refuses to be wrong about arithmetic. That is what makes the rest of it safe to automate.
The boundary is worth stating plainly, because it is where honest systems separate from confident ones. dagara proves your books hold together. It does not prove they match reality. Whether the invoice was genuine and the judgement sound are questions for people, served with evidence instead of sent to go and find it.
A transaction is the relationship between two accounts. The space between them is where the meaning lives, and it is the part every other system throws away.