Astin Consulting

Authorised to commit

business · professional services

Authority Basis

Domain verificationastinconsulting.com.au

ConformanceGUARD 2 — Enforced

AI Twinlive

ClaimedOwner claimed

Endorsements received2

Approved memberships2

What This Twin May Do

Be referencedRisk 1

Permitted

Appear as an AI-readable reference record that agents can find and read.

Every Presence in the Community is machine-readable by default.

Public facts only

Represent factsRisk 1

Permitted

State the entity's own information as its position (services, hours, location).

Domain is verified — facts are attributable to the entity.

Verified facts distinguished from self-declared

Answer questionsRisk 2

Permitted

Let the AI Twin respond to agent and human enquiries on the entity's behalf.

Verified entity with an active AI Twin.

Auto-revokes if conformance lapses or the domain is unverifiedMust not invent capabilities or give regulated advice

Accept & route enquiriesRisk 2

Permitted

Receive qualified enquiries and route them to the entity with provenance.

Verified, conformant (GUARD 1+) entity with an active Twin.

Auto-revokes if conformance lapses or the domain is unverifiedEnquiry provenance recorded

Make commitmentsRisk 3

Permitted

Issue quotes or accept bookings within owner-defined bounds.

Verified entity at GUARD 2 (Enforced) with a live Twin.

Auto-revokes if conformance lapses or the domain is unverifiedApproved partners / member Spaces onlyHigh-consequence actions require human approval (T3)

TransactRisk 3

Conditional

Enter agentic-commerce transactions — carts, receipts, WISE contracts.

Needs GUARD 3 (Provable) conformance.

Auto-revokes if conformance lapses or the domain is unverifiedApproved partners / member Spaces onlyHigh-consequence actions require human approval (T3)

AIH-12 horizon & AIT-12 threshold

Indicative · v0.1 drafts

Horizon

H5 · Agentic

Threshold

T5 · Delegated authority and interruption

Partially meets

Authority ceiling

Act within limits

Rung 5 of 7

The Twin is permitted to make commitments or transact on the entity’s behalf — that changes external state under delegated authority.

T1Purpose and intended use stated in writing

Twin description published on the Presence (self-declared). · Evidence: Assertion

T1Named operator and point of contact

Owner name and contact email on record (self-declared). · Evidence: Assertion

T1Basic usage logging

9 record changes on the hash-linked provenance ledger; individual Twin conversations are not logged there. · Evidence: Demonstration

T1A person decides what to do with every output

Twin answers are informational and enquiries route to the owner, but no review procedure is on record. · Evidence: Assertion

T2Data sources, memory and retention disclosed

Intelligence scope published; retention is not stated. · Evidence: Assertion

T2People are told when they are dealing with an AI system

The Presence page labels it “AI Twin” and the chat panel “Ask the Twin”, answering only from declared data. · Evidence: Demonstration

T2Personal data handled under a stated lawful basis

No lawful-basis statement on record.

T2Human review before output informs a decision about a person

Not on record.

T3Generated content labelled or carrying provenance metadata (R-EAL compatible)

Answers appear inside the “Ask the Twin” panel; no R-EAL provenance metadata is attached yet. · Evidence: Demonstration

T3Review step before publication or external use

Not on record.

T3Content and intellectual-property policy published

Not on record.

T3Periodic sampling of outputs against that policy

Not on record.

T4Reproducible evaluation suite for the intended tasks

Not on record.

T4Tool allow-list, even where tools are advisory

Permitted actions are bounded by the Control Plane capability model (a model, not runtime enforcement). · Evidence: Assertion

T4Intermediate steps logged and retained for a declared period

Not on record.

T4Known failure modes documented

Not on record.

T5F-ACT declaration covering Grant, Usage, Audit, Revocation and Data

GUARD level self-declared on the provenance ledger; a full five-control declaration is not on record. · Evidence: Assertion

T5Least-privilege credentials; declared and effective authority reconciled

Capabilities are scoped in the Control Plane; effective permissions are not independently reconciled. · Evidence: Assertion

T5Authority budget: spend, rate and scope limits

No spend or rate limits on record.

T5Human approval for irreversible or high-value actions

Required by the Control Plane model for high-consequence actions; not enforced at runtime. · Evidence: Assertion

T5Interruption and revocation tested, not just documented

No interruption or revocation test on record.

T5Tamper-evident action log

Hash-linked provenance chain verifies intact for this entity’s records; Twin actions are not individually logged. · Evidence: Demonstration

Honest statement

Indicatively classified at AIH-12 H5; partially meets AIT-12 T5 (v0.1 draft) — 19 of 22 controls not fully evidenced; evidence below profile E5; derived by Place & Space from self-declared and platform records.

Required evidence for this threshold: E5 · Reproducible test of interruption and revocation.

Worked out from this entity’s records by Place & Space. Not a certification, not a licence and not an OCC classification. For a formal self-classification use the open OCC classifier and read AIH-12, AIT-12 and AIE-12.

Standing Governance Rules

Capability ≠ authority — nothing here overrides the owner’s configured limits.
All authority is scoped and revocable, and lapses automatically if conformance or verification is lost.
Every grant, use and revocation would be written to the provenance ledger.

View in Society Graph View Presence Trust Report OCC Agent Record (JSON)Agent-readable JSON