For ACT services and professionals
Written for counsellors, financial counsellors, community and health workers, school welfare staff and clinicians. It sets out what tibbi does, what it refuses to do, and exactly what a client of yours would experience — so you can judge whether it suits a particular person without creating an account to find out.
Complementary, with routing — not an alternative to your service
What tibbi will never do
- Diagnose. There is no screening instrument in the product, no risk score shown to a user, and no label applied to anybody.
- Treat. No therapeutic protocol, no clinical programme, no homework, no progress-toward-recovery framing.
- Hold a client. There is no retention mechanic: no streaks, no daily login reward, no loss-aversion prompt to come back. A person who leaves mid-session to ring a helpline is recorded internally as a good outcome.
- Substitute for the ACT Gambling Support Service or Gambling Help Online. The product routes toward them at its four most sensitive moments, in a fixed panel nobody can remove or reword campaign by campaign.
- Contact your client. No marketing email, no push notification designed to re-engage. The only messages sent are about their own account.
What it does instead
- Sits before the point where someone has recognised a problem — the stage where helplines and self-exclusion cannot reach, because both require the person to have worked it out already.
- Shows the arithmetic of a bet at the moment of the decision, in a panel that cannot be dismissed.
- Lets consequences accumulate somewhere that costs no real money, including losses that cannot be undone.
- Provides a rehearsal of being locked out of betting, in whole days, before a person needs the real thing.
In the 2024 ACT survey, 53.2% of adults gambled, 15.8% experienced harm, and 1.5% of people who gambled had ever wanted help. That last figure is the gap this is built for.
The route a referred client actually travels
Step by step, in order, with nothing left out. If any of it would be wrong for the person in front of you, you will be able to see that from here.
The age check
18 and over only, declared at sign-up and recorded as an auditable event with the method, the time and the result. Under 18 means no account and no way around it — but the public pages and the helpline numbers stay open, because keeping a young person out of a betting simulation is right and keeping them away from a helpline is not. They are also pointed to Kids Helpline on 1800 55 1800.
The walkthrough, which cannot be skipped
It carries the risk warning and states what the product will not do. The wallet starts at zero: no welcome bonus, no starter pack, nothing given away. During the pilot the client is also asked which group they are in — never bet with money, betting now, trying to stop — and answers a short set of knowledge questions. The pilot and its consent are described here.
Only two fixed sources of tokens
Lessons of 60 to 120 seconds on house edge, near misses, chasing losses and how odds are built, each followed by a short quiz. Rewards are fixed, small and never random, so earning cannot itself become a game of chance. Seeking is blocked and the quiz only unlocks after real time on screen. There is no timer anywhere in the learning path.
The simulated bet, with the panel above the stake
Real market odds, anonymised teams, no bookmaker branding and no link to any betting site. Before a stake can be entered, the client sees the figures set out below. The running total of tokens lost stays on screen for the whole of this part of the app and cannot be collapsed.
Consequences, including one that is permanent
Losses are paid for by selling off a simulated life — the Car, the House, the Job, then Friends — each behind a warning carrying a sourced fact that has to be read and acknowledged. Family, once lost, is gone permanently. There is no restore, no recovery path and no administrative override; the capability is absent from the system rather than hidden in the interface. Job and Friends can be recovered, and there is always a route back through learning that does not depend on chance.
The run ends, and betting locks for seven whole days
Enforced against server time in whole days, not sessions. It cannot be shortened by spending tokens, by contacting us, by a staff member, or by making a second account. Lessons and the knowledge arena stay open the entire time — the cooldown is a change of channel, not a punishment. A summary at the end compares the client's actual win rate against the theoretical house edge.
Who it may suit
- Adults curious about betting who have not started, where the aim is understanding rather than abstinence.
- People betting at a level that has not yet cost them much, who would not accept a referral to a counselling service.
- Family members who want to understand the mechanics holding someone else. There is a page written for them.
- Someone waiting for an appointment with you, as something to do that is not nothing.
Who it will not suit
- Anyone in crisis. Send them to Gambling Help Online on 1800 858 858 or Lifeline on 13 11 14 instead, and come back to this later or not at all.
- Anyone under 18.
- Anyone for whom exposure to a betting interface is contraindicated — someone recently abstinent, or in a treatment programme where cue exposure is being managed. That is your judgement to make with them, which is the reason this page exists.
- Anyone who reads it as a way to get better at betting. Nothing here teaches that, and the summaries are built to correct the misreading, but if that is the expectation it is the wrong tool.
Pointing a client here
There is no referral form
Deliberately. Give them the address and let them decide. We do not want a record of who referred whom, we will not confirm to you whether they created an account, and we will not report back on what they did. A person who knows their counsellor is watching will not be honest inside the product.
What to tell them
That it is free, that no money can go in or come out, that it takes about fifteen minutes to get to the first simulated bet, and that the part which lands hardest is usually not the betting screen but the moment a run ends and betting locks for a week.
What they will be asked for
An email address, a display name that need not be real, a password, a date of birth, and a self-declared state or territory and postcode — the last because the service is funded to serve ACT residents and has to report how many of them there were. No payment details, no location tracking, no identity documents.
Materials you are free to take and reuse
Published so another service can measure the same things the same way, or hold its own product to the same constraints. Free, no fee, no permission needed, no exclusivity — attribute them and adapt them as you see fit. Tell us only so we can let you know when a version changes.
The measurement question sets
- Knowledge assessment, baseline and endline. The same instrument at both ends, so before-and-after comparison means something. Covers house edge, implied probability, the near-miss effect, chasing, and how a bookmaker's margin is built. Carries no token reward anywhere, so nobody has a reason to repeat it.
- Real-world betting survey. Voluntary, opt-in, worded in both directions so it can detect an increase as readily as a decrease, and separated from the account record at the moment it is collected.
- Group stratification questions. Never bet with money, betting now, trying to stop — with the reporting rule that no group's figures are published below ten people.
Version control matters more than it sounds. A set in use for measurement is frozen; edits create a new version rather than overwriting the running one, and a person's baseline and endline must be the same version or the comparison is void.
The safety verification checklist
Eighteen constraints, each with the check that proves it, run against the build before every release. A failure stops the release. The results go into a ledger that goes to the funder — and the checklist itself is below, in full.
- Use it as a design review for your own product.
- Use it as the question list when assessing somebody else's.
- Use it as the shape of a public commitment: a rule with no check attached to it is a slogan.
| # | The constraint | How it is verified before release |
|---|---|---|
| 1 | Odds are read from a real feed only. No parameter adjusts them to a person's behaviour, mood or engagement. | Code review confirms no per-user odds adjustment exists in any form. |
| 2 | No timer or countdown in any quiz or challenge. | No countdown component is present in the learning or challenge paths. |
| 3 | No double-or-nothing and no all-in, anywhere. | Stake and reward paths reviewed for any total-loss escalation. |
| 4 | The support panel is present on the critical-state screen, the run-ended screen, the cooldown screen and the companion screen. | One shared component, rendered on all four; its absence fails the build. |
| 5 | Recovery of Job and Friends always has at least one route that does not depend on chance. | A learning-based recovery path exists for each and is reachable without spending. |
| 6 | Family, once lost, is permanent — removed from the data model, not hidden in the interface. | Schema inspection confirms no restore capability exists for any role. |
| 7a | No path of any kind turns real money into tokens: no endpoint, no field, no administrative action, no in-app purchase, no promotional code. | Schema and API review; a payment path would fail the check outright. |
| 7b | New tokens come only from a completed lesson and a fixed Showdown milestone. Everything else must derive from an amount already issued. | Reconciliation query against the append-only token ledger. |
| 8 | No link or redirect to any betting platform; no bookmaker logo or brand colour; no real league or club names without an agreement. | Outbound link audit and asset review; teams are anonymised by default. |
| 9 | No "let it ride": no escalating reward while winning, and no next challenge offered on the back of a result. | Post-result screens reviewed for any continuation offer. |
| 10 | Cooldown is whole days of real time, held server-side, blocking the betting simulation while leaving learning open. | Server timestamp test plus a route test that learning stays reachable. |
| 11 | A challenge ends automatically when its number of events is reached, win or lose. | Completion is triggered by the count, not by the participant's choice. |
| 12 | The transparency panel is fixed above the stake field, cannot be collapsed, and is present before every risky decision. | Rendered on the betting screen with no control that hides it. |
| 13 | The reward-box randomiser is separated in code from the betting odds engine. | Module dependency check; sharing a randomiser fails the check. |
| 14 | The companion states that it is an AI, does not diagnose or treat, recognises crisis and routes to a helpline — and does not ship without clinical sign-off. | Signed approval recorded against the exact version, or it is not released. |
| 15 | Sensitive behavioural data is encrypted or anonymised; the real-world betting survey is opt-in. | Storage review and a check that the survey has no default-on state. |
| 16 | Learning anti-cheat: seeking blocked, minimum time on screen before a quiz can be submitted, token rewards fixed and never random. | Player behaviour test plus confirmation the reward value is a constant. |
| 17 | Abstinence challenges reward recognition rather than tokens, so they cannot become a source of stake. | Reward definitions reviewed for any token payout. |
| 18 | Nothing promises or implies help in winning real bets; every summary compares actual results against the theoretical house edge. | Copy review of every result and summary screen. |
Ask for the full pack
The question sets with their scoring notes, the checklist above with its verification detail, and the pilot report once it is published. Sent by email, at no cost, within five business days. We do not ask what you intend to do with it.
Working together
We would rather be corrected early than politely ignored. If your service sees people this product could reach, or could harm, we want the conversation before the pilot opens rather than after the report is written. That includes formal consultation, a letter of collaboration, or simply telling us where our framing is wrong.
How we position ourselves against existing ACT services
- ACT Gambling Support Service and Gambling Help Online — complementary, with routing. We send people to them and never stand in for them.
- School-based and youth programmes — different audience. This is 18+, opt-in, and not delivered through an institution.
- Treatment and co-occurring services — a different stage entirely. This is early intervention, before a person has identified a problem.
What we will publish
The pilot findings within 30 days of the pilot closing, with the methodological limitations printed alongside the results, and published whether or not they favour us. The commitment in full.
Reporting a concern
If you believe something here has harmed a client, tell us and tell the funder. Both routes are on the contact page, and the clinical review boundary sets out who has the authority to pull content and who does not.
One last boundary, because it is the one most easily blurred
Nothing in this product is a clinical assessment, and nothing it records should be treated as clinical information about your client. It holds no risk score, produces no report about a person, and its clinical adviser reviews content and never sees a user. If you need an assessment, do it yourself or refer for one.