Autonomy level
NVIDIA's classification framework for how much independent action an agentic system can take, cited and adopted by the CSA Addendum (pp.18 to 20), ranging from a Level 0 system making straightforward inference calls to a Level 3 system that can dynamically modify its own execution paths. This is a distinct scale from the MGF's four levels of human involvement: autonomy level describes how independently the agent's own workflow can branch, while human involvement describes how much a human reviews before or after it acts, and a deployment's position on one scale does not determine its position on the other. A higher autonomy level generally requires more granular, more frequent monitoring, because more of the agent's behaviour is determined at runtime rather than fixed in advance.
Defined in 2 GAGE programs, which carry 2 distinct definitions of it. The wording above is taught in Certified Agentic AI Governance Professional (CAAGP).
How each discipline defines it
The same term does different work depending on who is using it. These are the definitions as each program teaches them, unedited.
NVIDIA's classification framework for how much independent action an agentic system can take, cited and adopted by the CSA Addendum (pp.18 to 20), ranging from a Level 0 system making straightforward inference calls to a Level 3 system that can dynamically modify its own execution paths. This is a distinct scale from the MGF's four levels of human involvement: autonomy level describes how independently the agent's own workflow can branch, while human involvement describes how much a human reviews before or after it acts, and a deployment's position on one scale does not determine its position on the other. A higher autonomy level generally requires more granular, more frequent monitoring, because more of the agent's behaviour is determined at runtime rather than fixed in advance.
How much a tool acts without per-action human review, on a spectrum from suggest (AI proposes, human does), to approve (AI drafts, human approves each), to exception (AI acts, human handles flagged cases), to full (AI acts with monitoring only).
Where it is taught
The exact lessons this term appears in. The first module of every program is free with a free account.
- Using AI Tools: Automation, Chatbots, and Analytics · Practical AI Workflow Design and Prompt Engineering, Certified AI Practitioner: Workplace Foundations
- Watching an Agent Work · Rollout, Monitoring, and Incident Response, Certified Agentic AI Governance Professional (CAAGP)
Where it shows up at work
A definition is where Autonomy level starts. The lessons above teach it inside the competencies below, which are what job descriptions ask for and what the GAGE credential records. Every link here is drawn from the same careers map that builds a role's readiness check.
Writes requirements, runs a fair evaluation, pilots within limits, and refuses a demo as evidence.
One lesson on this page teach it as a primary outcome.
Decides where AI should be used, where it should not, and which use cases create measurable value, tied to the organization's priorities.
One lesson on this page teach it as a primary outcome.
Sets performance metrics, thresholds and review triggers after launch, and treats a model change, a vendor update or new data as a reason to re-check.
One lesson on this page teach it as a primary outcome.
Roles that hire for it
Each role page shows the full competency set, what the seat decides, and a readiness check against your own evidence.
- Chief Information Officer, technology leadership editionneeds it at depth
- AI Vendor Risk Managerneeds working knowledge
- Chief AI Officerneeds it at depth
- Chief Technology Officer, AI engineering editionneeds working knowledge
- Third-Party AI Risk Analystneeds working knowledge
- AI Product Manager, responsible product editionneeds working knowledge · 3 open postings this week
How job descriptions say it
Postings rarely use the textbook word. These are the phrases the careers map reads as asking for the competencies above.
Where the frameworks place it: CISA software acquisition guidance; EU AI Act Article 72; NIST AI RMF MEASURE 3 and MANAGE 4.
One interview question about it
How do you buy an AI product well, from writing the requirement to the questions you ask in the demo?
Asked of candidates for buying ai well. Answering it in your own words is exactly how a GAGE topic is passed, and the credential records the pass where an employer can check it.
If you come from another field
Some backgrounds already carry this competency, and the readiness check gives credit for it.
- A business, MBA or technology management degree carries it as a claim to back up: Strategy and opportunity triage are the core of the degree.
- Model risk, credit risk and quantitative analysis carries it as a claim to back up: Performance monitoring thresholds are familiar.
Terms it appears with
Not an alphabetical neighbourhood: these are the terms taught in the same lessons, ranked by how often they appear together.