How the score works.
AI-native is an operating model where agents do the work and people make the calls. This assessment scores the model. The tools you pay for score nothing.
What gets scored
14 questions, two for each of seven dimensions. Each answer scores 0 to 3, so a dimension scores out of 6 and the total out of 42. The answers describe real conditions in a company, and the one that is true is often the one you would prefer not to pick.
- Workflow legibility
- Before a workflow can be automated, it has to be made legible.
- Context readiness
- An agent only knows what your company has written down.
- Data and systems
- Agents need one answer to every question about a customer, an order or a ticket.
- Human–agent boundaries
- The human-agent boundary is the architecture.
- Production controls
- A system that cannot be inspected cannot be trusted.
- Adoption and ownership
- Every workflow needs an owner with the authority to change it.
- Business impact
- If no number moved, it was theatre.
The five levels
Your total places you on one of five levels.
| Level | Total | What it looks like |
|---|---|---|
| AI Tourist | 0–8 | Occasional tool use. The operating model is untouched. |
| AI-Curious Team | 9–16 | Subscriptions and prompts. Activity without structure. |
| AI-Assisted Team | 17–25 | Real individual gains. Nothing the business can build on yet. |
| AI-Native Operators | 26–34 | Repeatable workflows owned by teams. The foundation exists. |
| AI-Native Company | 35–42 | Core operations redesigned around controlled AI systems. |
The two caps
A total can hide a serious gap, so two rules limit how far it can carry you. When one applies, the result says which and why.
- Zero on production controls holds you at AI-Assisted Team.
- A system that cannot be inspected cannot be trusted, so a high total cannot carry a company past AI-Assisted Team.
- A dimension scoring 1 or below holds you at AI-Native Operators.
- Autonomy scales on structure, and an average hides the weak part. The top level needs every dimension to hold.
Your weakest link
The result is built around your lowest-scoring dimension. Ties go to the one that would do most damage in production, with production controls first. The diagnosis quotes the answers you gave, so you can check it against what you said.
What it cannot tell you
This is a self-assessment, so it measures what you tell it. Agent mode is the harder test, because an agent scores zero wherever it cannot see the evidence. There is no comparison with other companies yet, because that needs data this site does not collect.
It also cannot tell you which of your operations is worth redesigning first. That is the work of a Blueprint, Gris Noir’s fixed-fee diagnostic.
Fourteen questions. What would yours score?
Three minutes on how your company really runs. No email, and your answers stay in your browser.
Take the assessment