Behavioural Governance for AI™

Prove oversight happened. At the moment of decision.

Businesses can prove they trained their people on AI governance. Few can prove meaningful human oversight occurred when an AI-assisted decision was made. TrustShyft™ is being built to close that gap, with oversight records captured at the moment of decision. Append-only. Recoverable. Ready for scrutiny.

Talk with us Explore the framework

Set boundaries
Ensure accountability
Drive consistency
Build trust

The evidentiary gap

AI governance rarely fails from lack of intent.

It fails because AI changes the conditions inside a business faster than frameworks adapt. Policies get written. Owners get named. Training gets completed.

Then a regulator, a court or a board asks a simple question. Who reviewed this decision, against which policy, and what did they see? Most businesses can only answer by reconstructing events after the fact.

Reconstruction is not evidence. Evidence is created at the moment the decision is made.

The framework

Behavioural Governance for AI™

Governance does two jobs. Most frameworks only do the first. Behavioural Governance for AI™ treats the second as a design responsibility in its own right.

Direction-setting

Policies, principles and named owners. It tells people what should happen. It is necessary. On its own it is not enough.

Condition-setting

The environment inside which governance must hold. Workload, tooling, incentives and decision design. This is where oversight succeeds or quietly fails.

Design accountability for conditions sits at the operating model level. Not at the board. Not in the policy.

Read the newsletter series

The platform · in development

TrustShyft™ is designed to bind evidence at capture time.

Each oversight record is created when the decision is made and bound to the policy version in force at that moment. Records cannot be edited or deleted through the platform.

Resolution-time binding

Reconstructed after the fact

Evidence assembled when someone asks
Policy version inferred, not recorded
Gaps look like absence of action

Capture-time binding

Recorded at the moment of decision

Evidence created as the decision happens
Bound to the active policy version
Failures and rejections are recorded too

Append-only records

Once saved, entries cannot be edited or deleted through the platform. Corrections are added as linked entries, so the original stays visible.

Policy version binding

Each record resolves to the exact policy that applied when the reviewer acted.

Failure states as evidence

Rejected actions and invalid modification attempts are governance evidence. They are kept, not discarded.

Recoverable on request

When a regulator or board asks, the answer already exists. No reconstruction required.

Built with today's obligations in view

Regulators are moving from asking whether a business has AI policies to asking for evidence those policies operated.

Privacy Act APP 1.7 · from 10 December 2026
APRA CPS 230
ASIC REP 798
EU AI Act
ISO 31000
Robodebt Royal Commission findings

Founder

Sandra Farquhar

Sandra is a senior business analyst with more than 20 years of delivery across justice, rail, retail and regulated SaaS. Her work sits where policy meets operating reality.

She created Behavioural Governance for AI™ to answer the question her delivery work kept raising. How does a business prove its people actually exercised judgement when AI was in the loop?

She applies the same discipline to construction through Programme Shyft™. TrustShyft™ is based in Brisbane and works with businesses across Australia.

Connect on LinkedIn

Ways to work with us

The framework, applied to your workflows.

Every engagement is scoped to your sector, your obligations and the decisions that matter most. Start where the risk is highest.

Assessment

One AI-assisted workflow, assessed against the framework and scored with the JRI methodology. You see where oversight holds, where it fails and what to change first.

Talk to us about scope

Platform

TrustShyft™ is in development. Talk to us about shaping it around your workflow, with oversight records captured at decision time and evidence you can produce when it is asked for.

Talk to us about scope

Sector implementation

The framework tailored to your sector's obligations and operating model. Conditions designed so oversight holds in practice, not only on paper.

Talk to us about scope

Advisory

Ongoing guidance for boards and risk functions. Briefings, workshops and talks on behavioural governance for AI.

Talk to us about scope

Get in touch

Start where the decisions are real.

TrustShyft™ works with businesses using AI in decisions that carry regulatory, customer or reputational weight. Tell us where oversight matters most in your business.