AI readiness checker
Check your store’s information and ownership, then download the actions still needed.
Your self-assessment
Actions to complete
Checks confirmed
6 of 9
Checks answered
8 of 9
This is a self-assessment of your answers, not compliance certification or a guarantee that model answers are correct.
6 of 9 checks confirmed. 1 unanswered. 3 actions to review.
Readiness by area
| Area | Your answers |
|---|---|
| Data | 2 of 2 confirmed; 0 unanswered |
| Policy | 1 of 1 confirmed; 0 unanswered |
| People | 0 of 2 confirmed; 1 unanswered |
| Controls | 3 of 4 confirmed; 0 unanswered |
Your next actions
Named agent owner: No
Appoint an accountable agent owner with authority to organise fixes. Put this arrangement in place. A named owner makes it clear who handles problems and approves changes.
Handover to a person: Not answered
Define handover triggers, a receiving person or team and the out-of-hours response. Confirm the current position, then complete this action if needed. A shopper who needs a person needs a working route to help.
Answer testing and review: Partly
Finish the answer test set and assign recurring reviews with a correction process. Finish the remaining work. Complete source information alone cannot prove that generated answers are correct.
Every statement is confirmed by you. Keep testing and reviewing the agent. These answers do not approve a launch.
What the readiness checks measure
The AI readiness checker turns your answers into a preparation list. It checks what you report about your store, without inspecting your website or testing an agent.
Data covers complete product records and a working price and stock update process. Policy checks whether the agent has usable shipping and returns facts. People covers an accountable owner and a handover route. Controls covers answer review, human intervention, personal-information handling and disclosure to shoppers.
- Product descriptions, variants, sizes and relevant specifications are complete and checked.
- Published policies state delivery areas, charges, timeframes and returns conditions. The agent can use them.
- Staff know when to take over, who receives the conversation and what happens outside staffed hours.
- Staff test shopper questions before launch and review answers after launch, with a process for correcting failures.
- A trained person can pause the agent or stop an unsuitable action and keep customer help available.
- You have reviewed what personal information the agent uses, why it is needed, who receives it and what shoppers are told.
For example, a shipping policy that omits remote delivery areas cannot support a reliable answer about delivery to those areas. Add the missing rule to the policy, then test the agent’s answer. A product catalogue cannot supply that policy fact.
Each Yes adds one confirmed check in its area. Partly and No count as answered but remain on the action list. Not answered stays visible and asks you to establish the current position. Counts are not weighted, and there is no passing percentage.
The question set and action order are editorial choices informed by Australian Government AI adoption guidance and the OAIC checklist for commercially available AI products. They are not an official assessment scheme.
Only Yes counts as confirmed. Partly, No and Not answered each keep an action on your list.
Choose Yes only when the whole statement is true. Choose Partly when work remains. Leave anything you have not checked unanswered.
Turn the result into a launch list
A hypothetical store has complete product records, current price and stock updates and published shipping and returns policies. It has human intervention, personal-information handling and AI disclosure in place. Its agent owner is No, handover is Not answered and answer review is Partly.
Load the example store to see 6 of 9 checks confirmed and 8 of 9 answered. Data shows 2 of 2 confirmed, policy 1 of 1, people 0 of 2 and controls 3 of 4. The single unanswered check is handover.
The next actions are to appoint an owner, establish who receives escalations and finish the answer-review process. Clean product data leaves those staffing and review tasks unresolved. Agree who receives an unsupported returns question, what happens outside staffed hours and who corrects a wrong answer.
Download the assessment and assign a person and completion date to each remaining action. Test the handover and corrected answers before changing those checks to Yes.
Download a text file with your answers and actions. Answers stay in this page and are cleared when you reload.
Start with ownership, personal-information handling and human intervention. Then work through handover, disclosure, store information and answer review. Adjust this suggested order to your store’s risks.
Readiness and AI governance
This self-assessment records your preparation. It cannot certify privacy compliance, prove catalogue accuracy or guarantee correct model answers. Even nine Yes answers need evidence from your store and testing of the agent.
Australian Government guidance calls for accountability, testing and ongoing human oversight. The OAIC checklist asks organisations to examine how AI handles personal information and how staff check its accuracy. The controls here help identify work in those areas.
Use the AI governance guide to set the wider ownership and review rules. Keep this checklist with your launch tasks, and repeat it when your source information or the agent’s responsibilities change.