Agentic Trust Index

How much authority does your AI deserve?

ATI evaluates whether the identity, controls, boundaries, and evidence surrounding an intelligent system justify the authority it has been granted.

Assess Your System Explore the Methodology

ATI Screen v1.0 Beta20 questions · 7–10 minutes · No account · Processed in your browser

ATI Trust Profile · Example

Illustrative

Customer-support reply agent (pilot)

/ 100 trust readiness

Authority exposure: High
Evidence confidence: Moderate
Current authority (reported): Bounded Autonomy

Recommended authority boundary

Act with Approval

Human approval required for consequential actions.

⚠ Authority exceeds assurance: the system holds more authority than its evidence supports.

Generated by the ATI Screen engine from a sample answer set. Not a real system.

The problem

Capability is not permission.

The question is no longer simply what AI can do. It is what AI should be authorized to do, and what evidence should be required before that authority is granted.

AI systems are increasingly capable of:

access sensitive information
invoke tools
modify systems
communicate externally
execute transactions
delegate tasks
make consequential decisions
…often without anyone deciding they should.

The ATI question

Not “is this AI safe?” but “what authority is justified?”

ATI separates two things that are usually blurred together, and judges one against the other.

Authority exposure

What the system can do, and what happens if it is wrong.

Actions available, data reachable, blast radius, impact of failure. Exposure describes stakes; it is not a judgment of quality.

judged against

Assurance strength

What constrains it, and what proves the constraints hold.

Attributable identity, deterministic enforcement, enforced boundaries, and evidence that survives a skeptic.

When exposure outruns assurance, ATI says so plainly, and recommends the authority boundary the evidence actually supports.

The ATI model

From self-report to independent assurance.

Each level asks more of the evidence. ATI Screen is available today; the rest of the model is where ATI is going.

01
Available
Screen
Self-reported
A 20-question self-assessment that establishes authority exposure, control readiness, evidence confidence, key gaps and a recommended authority boundary.
Start ATI Screen
02
Coming Soon
Verify
Evidence-supported
Technical verification of the evidence behind a Screen result: identity and permission configurations, tool definitions, policies, logs and approval mechanisms.
About ATI Verify
03
By Request
Assure
Independently reviewed
Expert-led, independent evaluation for production agents, vendors and high-impact deployments, where self-assessment is not enough.
Request an assessment
04
Research Program
Benchmark
Comparative intelligence
Aggregated, anonymized, opt-in research on how organizations constrain and evidence autonomous systems. Data collection underway.
ATI Research

Screen → Verify → Assure → Benchmark. Only ATI Screen is automated today. ATI does not currently analyze uploaded artifacts.

Five domains

The questions every consequential system should be able to answer.

Domain 01

Authority

What can the system actually do?

The actions available to the system, the consequences if they go wrong, and whether consequential actions pass through an approval that is bound to the specific action.

Domain 02

Identity

Under whose authority is the system acting?

Whether actions are attributable to a distinct, scoped, non-human identity, and whether delegated authority narrows as it moves between users, agents and tools.

Domain 03

Enforcement

What actually prevents the system from exceeding its authority?

The difference between soft controls that guide behavior and deterministic controls that execute outside the model, on every action, and fail closed.

Domain 04

Boundaries

Where does the system's authority stop?

The edges of what the system can reach: data, tools, destinations, environments, tenants and downstream dependencies.

Domain 05

Evidence

Can the organization's trust claims actually be demonstrated?

Whether claims about the other four domains are backed by attributable records, tests, dated artifacts and an accountable owner.

Evidence maturity

A claim is not a control.

ATI tracks how far each trust claim has travelled up the evidence ladder, and limits what a result can say accordingly.

01
Assertion
Someone states that the control exists.
02
Documented
A design, policy or procedure describes the control.
03
Implemented
Current configurations or records show the control is in place.
04
Tested
Dated test results show the control operates, including attempts to bypass it.
05
Independently Verified
A qualified party outside the delivery team examined the evidence.

Find out whether your system’s authority is justified.

Twenty questions. A trust profile, the gates that apply, the controls that matter most, and the evidence that would raise confidence.

Assess Your System Need independent review? Request an Assessment