Every price is here, in Australian dollars excluding GST. They are founding prices, and we review them after the first three studies.
Frame
Where should our AI act on its own, and where should it ask?
A$7,500Fixed / Half a day, written map within a week
A shared map of act, ask and stay quiet for your feature
The three questions most worth testing
A recommendation on which study to run first
What the price assumes and what changes it
Assumes
One half-day session with your team, one feature in scope, one written map. AI drafts the narrative and the first analysis. A named specialist signs off every technique.
Changes it
A second feature, a second team, or a session outside Sydney.
One system, with its documents and access to the people who built it. AI drafts the narrative and the first analysis. A named specialist signs off every technique.
Changes it
More than one system, or a system we cannot see inside.
How much autonomy will our users trust, and where will they take control back?
A$24,000Fixed / 4 to 6 weeks
A map of every consent moment
Gaps ranked by how likely they are to fail
Redesigned notice and consent surfaces, tested with people
What the price assumes and what changes it
Assumes
One system and one group of people. The number of sessions is agreed before we start. AI drafts the narrative and the first analysis. A named specialist signs off every technique.
Changes it
A second system, further user groups, or a second round of redesign.
A threshold map: when it was noticed, when it was missed
A go, change or stop recommendation
What the price assumes and what changes it
Assumes
One behaviour, one setting, and an off-the-shelf or kit device standing in for the final object. Participants and sessions are agreed before we start. AI drafts the narrative and the first analysis. A named specialist signs off every technique.
Changes it
A second behaviour, a second setting, a longer placement, or custom hardware. Custom hardware moves you to the device study.
Can we see it working in real rooms, with real hardware?
From A$100,000Scoped / Scoped per project
Working prototypes, from scale models to a device in use
A bill of materials with certification routes
Evidence from real rooms and real people
What the price assumes and what changes it
Assumes
Your study says go. Hardware, tooling, supplier fees and certification are passed through at cost and listed in the scope. AI drafts the narrative and the first analysis. A named specialist signs off every technique.
Changes it
Custom radio design, the certification route, the number of devices and the number of rooms. A full-scale custom device changes scope, price and timeline. We quote any custom radio work separately before you commit.
The workshop, the map, the review and the study carry fixed prices. The assumptions are written on each card.
From, where hardware decides
The device study starts at A$100,000. Hardware, tooling, supplier fees and certification are passed through at cost and listed in the scope.
What moves the price
Each card says what changes it. A second behaviour, a second setting or custom hardware does. We tell you before you commit.
Who does the work
AI drafts the narrative and the first analysis. A named specialist signs off every technique. We show effort as weeks and checkpoints, not hours.
The answer does not move the price
Findings are reported as found, including stop. The fee is the same for go, change or stop.
What this is not
Design work. It is not legal advice and not an attestation. It sits alongside your counsel and your assurers.
What a verdict looks like
Every study ends in one page. This is the format.
Illustration only. An invented feature, not a client result.
The question
Should the room dim and lower the speaker when it senses a call starting?
What people noticed
The dimming, every time.
What they missed
The speaker lowering. Most people thought the caller had gone quiet.
The threshold
Dimming: act. Speaker: ask. Calls from numbers the person does not know: stay quiet.
Verdict
Change. Keep the dimming. Ask before lowering the speaker. Do not act on unknown calls.
Questions we expect
Is this only a privacy-policy update?
From 10 December 2026, organisations covered by the Privacy Act must describe significant automated decisions in their privacy policy. That is a disclosure rule, and a policy paragraph can meet it. It does not tell you whether anyone noticed when your system sensed them or acted for them. We test those moments with real people.
We already have a consent platform, or an assurance firm.
Those show that a control exists. We find out whether a person noticed it. The two sit side by side, and our findings give your platform and your assurers something to check against.
Can a small group of people prove anything?
No, and we do not claim a rate. A small group shows where a behaviour breaks, which is what a go, change or stop answer needs. The number of sessions is agreed before we start and written on the scope.
Why not ask a law firm?
Ask one as well. We are a design studio. This is design work, not legal advice and not an attestation. We send you back to your counsel with evidence they did not have.
Ten weeks is long when agents ship weekly, and our agent is software only.
The Workshop and the Consent Map take days and weeks, so you can start there. The study answers a question that outlasts a release: how much autonomy people will accept, and where they take control back. If your agent is software only, the study uses a screen and an off-the-shelf device in place of a custom object.
Who carries the risk if the answer is wrong?
You do. We report what we saw, including stop, and the price does not depend on the answer. We do not sell assurance. The decision stays with you and now has evidence behind it.
Start here
The quickest way in is the workshop. Or take the worksheet and run it yourself first.
