AI that thinks in Australia
AI is arriving in every membership platform. The only question that matters is where your member data goes when it does. Nexy's answer: nowhere. All AI on the platform is moving to sovereign Australian infrastructure, and the requirements any provider must meet are published here, before any partner is named.
Does AI in membership software send our data offshore?
It usually does. Most platforms add AI through an offshore API, so member records cross a border into another company's systems. Nexy publishes five requirements any AI provider must meet before it is named: inference in Australia, no training on member data, nothing retained, platform roles and audit inherited, and a client-controlled off switch.
Requirements first. Partners second.
Most vendors announce an AI partner, then explain the terms. We publish the terms, then sign the partner. These five requirements are the contract bar for any provider carrying AI on Nexy, and they are not negotiable in either direction.
Whoever we name will already have cleared this bar. You will be able to check.
Where does the AI think? Ask every vendor that.
The common way to add AI to software is an offshore API: member questions, member records and drafts crossing a border into another company's systems, under another country's rules. It works, and it is fast to ship. It is also a sovereignty decision made on your behalf, usually without being mentioned.
Nexy is taking the other road. AI on the platform today already operates under strict no-training terms, and every AI workload is moving to sovereign Australian infrastructure: inference onshore, nothing retained, nothing learned from your members, all of it governed by the same roles and audit as the rest of the platform. Slower to arrange. Worth it permanently.
The same governance. A new capability.
Inference onshore
AI processing on sovereign Australian infrastructure, so member data does not cross a border to be useful.
Never trained on
Member data never trains a model, ours or anyone else's. Not as a setting. As a term of contract.
Nothing retained
No prompt caching, no stored inputs or outputs beyond what your organisation keeps on its own record.
Governed and audited
AI inherits the platform's role-based access and audit logging. What it reads and does is what your roles allow, on the record.
You hold the switch
On or off is your organisation's decision, per capability. AI is something you govern, not something that happens to you.
Named when signed
Our infrastructure partner will be named the day the agreement is signed, published against these requirements, line by line.
Sovereignty is the moat, not the feature list.
Every platform will have AI features; within a year they will look much the same. What will separate vendors is the question boards are only beginning to ask: where does it think, and who learns from our members? Nexy's answer is structural, published and contractual. The full data sovereignty position sits here, and both pages are written to be taken into procurement.
Written to answer the questions your framework asks.
Boards are being handed AI governance frameworks: Australia's AI Ethics Principles, the Voluntary AI Safety Standard, and sector guidance built on top of them. Our requirements were written to be answerable against that language rather than beside it. We claim alignment, not certification, and we will answer the mapping in writing for your own framework.
| What the framework asks | What Nexy commits to |
|---|---|
| Privacy protection and security | Member data hosted in Australia, never used to train a model, inference moving to sovereign Australian infrastructure, nothing retained by the provider. |
| Transparency and explainability | Answers are grounded in your own records and content, traceable to source. The provider requirements are published before a provider is named. |
| Human-centred values and human oversight | AI is assistive, not decisive. Decisions affecting a member's standing, entitlements or money are made by your people. |
| Accountability | AI inherits role-based access and audit logging, so what was asked and what was done sits on the record with everything else. |
| Contestability | Because AI drafts and surfaces rather than decides, an output can be checked, corrected or ignored before it affects anyone, and the correction is logged. |
| Reliability and ongoing monitoring | Governance is a running system rather than a signed statement: continuous logging, role review, and a client-controlled switch to turn capability off. |
| Vendor and supply-chain risk | The platform is not built around one provider. Requirements are contractual, published first, and designed so the provider underneath can change without your processes being rebuilt. |
Asked before the first call.
Which AI provider will Nexy use?
Does AI on the platform today train on our data?
Will our member data leave Australia to use AI?
Will AI be optional?
Does this align with the AI governance framework our board has adopted?
Why publish requirements before naming a partner?
Take both pages into your next vendor meeting.
Ask where the AI thinks, who learns from your members, and who holds the switch. Then compare the answers with ours. The argument behind the product is in Issue 7 of The Platform Realist.