TLDR: An ontology is a structured model of how a business works: the kinds of things in it and how they relate. Chief Staffer starts every business from one General model of business operations, then authors your own model from what your business already produces. With a model, your AI knows what to look for instead of treating your files as plain text.
When people first hear that Chief Staffer uses an "ontology," it sounds like jargon. It is not. An ontology is just a structured model of how a business works: what kinds of things exist in it and how they relate. This post explains what that model contains, why Chief Staffer starts every business from the same General model, and how your own model is authored on top of it.
What an ontology is
If a knowledge graph is the record of your specific business (your clients, your projects, your deadlines) then an ontology is the template that record is shaped by. It defines the categories.
An ontology says things like: a business has clients; clients have projects; projects have deliverables and deadlines; people hold roles; money moves through invoices and retainers. It does not contain your actual clients. It describes the kinds of things a business has, the relationships that can hold between them, and the documents a business produces.
This matters because structure has to come from somewhere. Without a model of the business, an AI treats your files as undifferentiated text. With one, it knows what to look for and where each fact belongs.
The evidence for giving an AI a model
This is not only our opinion. On 3 June 2026 Anthropic published How Anthropic enables self-service data analytics with Claude. Anthropic pointed its own assistant at its own business analytics. The same AI went from answering 21% of real business questions correctly to above 95%. What changed was not the model. It was given a working map of the business and the skills to use it. Piling in more examples barely moved the score.
That is a result about the general approach, published by someone else. It is not a measurement of Chief Staffer. It is the reason the structure comes first.
Two parts: the General model and your own
Chief Staffer's model has two parts, and both are real.
The General model. Every business starts from the same one. It is a model of a whole corporation: what each department is accountable for, what each role does, the work each one performs and the documents each one produces. It spans 19 departments, 155 staff roles, 1,341 skills and 1,223 document templates, and it is modeled on the best open-source data and public research available. It is the floor a business starts from, not the ceiling. You can browse its departments, roles, skills and document template names through the Chief Staffer server, using the AI you already use.
Your own model. On top of the General model sits a model of your business, authored from your own material: your roles and who holds them, who you deal with, the sequences you run and where someone has to sign off, the documents you produce, the words you use and the rules you follow. It is private to you. Nothing from any business's own model is shown to anyone else, and the General model is the same for every business.
How your own model gets authored
Authoring starts with discovery: learning how your business actually runs from what it already produces, such as documents, transcripts, email, calendar and files. Discovery looks for structure, and it includes a gap analysis that maps your business onto the General model, showing what you already cover and the roles and work you have never been in a position to staff.
Your model is corrected by you before anything is authored from it. You can review it, correct it in a sentence, and check it. It stays yours.
Why start from a General model
You never start from nothing. The General model already understands how a business runs, so your own model adds specificity instead of rebuilding the basics.
It recognizes your real objects and documents. Your own model carries the specific roles, documents and rules of your shop. The General model gives them a shape to attach to, so the work starts from the right structure rather than a blank page.
It speaks your vocabulary. Your own model holds the words you actually use, so the AI you already use understands you with less translation and its output reads like it came from inside your business.
The General model is the floor, and it is a real floor. Your own model is the difference between an AI that understands business in general and one that understands the business you are actually in.
Read next: how the knowledge graph works, the record built on this model, and what an AI chief of staff is. For what the model covers across a business, see capabilities.