AI that drafts. People decide.
AI can take hours of searching, summarising and drafting off your team. It shouldn't take decisions away from them. Everything we build follows the rules we set for Centra AI in our own platform.
Our rules
Four rules, built in.
- Approval on every action
- Nothing is sent, saved or changed until a person says yes.
- Irreversible actions stay human
- Our AI can't post, dispatch, approve or delete. That's by design, not a setting.
- Answers from your own data
- It works from your live records, not a report someone ran last month.
- Open about where AI is used
- People can see when a draft came from AI.
ChemCo NZ
| BioGuard Disinfectant 5L × 65 | $1,852.50 |
| SaniWipe Canister 200 × 24 | $297.60 |
| Total | $2,150.10 |
Waiting for your OK
What we build with AI.
Practical tools that save time on work your people already do, with a person in control of the outcome.
Answers from your records
Ask a question in plain English and get an answer drawn from your live data.
Summaries
Long cases, documents and histories condensed into what a person needs to know before they act.
Early warnings
Exceptions flagged while there's still time to act: stock running low, invoices overdue, customers going quiet.
Drafts for approval
Emails, orders and responses prepared and waiting for someone to check, edit and send.
Triage
Incoming requests sorted and routed to the right person, who makes the call.
Inside your systems
AI added to the software you already run, through the integrations we build.
How we keep it safe.
Safety is designed in, not added on. These are the controls we put in every AI feature we build.
- An approval step before any action leaves the system
- Hard limits on what the AI is allowed to do at all
- A record of what the AI proposed and who approved it
- Testing against real examples from your work before release
- Clear labelling, so people know when they're reading AI-drafted content
- Your data handled under the same privacy and security standards as the rest of the system
The Public Service AI Framework.
Released by the Government Chief Digital Officer in January 2025, the framework sets out principles for safe, responsible AI in the public service. Here's how our approach lines up.
| Principle | What it asks for | How we approach it |
|---|---|---|
| Inclusive, sustainable development | AI that supports inclusive growth and doesn't widen inequalities. | Accessible interfaces to WCAG 2.2 AA and plain language throughout. |
| Human-centred values | Respect for the rule of law, human rights and privacy. | Privacy built in from discovery, and people make every decision. |
| Transparency and explainability | Being open about when AI is used and how it affects outcomes. | AI drafts are labelled, so people always know when AI was involved. |
| Security and safety | Robust risk management and traceable data. | Hard limits on AI actions, testing before release and independent penetration testing. |
| Accountability | Oversight, with people accountable for AI use. | A record of every AI proposal and the person who approved it. |
See responsible AI working in production.
We'll walk you through Centra AI in our own platform, and talk through where AI could help your team.
Ask for a walkthrough