We use AI where it does the job better than what you have now, and a person stays accountable for everything we deliver. Your code can go into AI tools; your business data goes in only with your approval. We only use AI tools that do not train on what we send them.
Effective October 7, 2026
01AI where it earns its place
Putting AI into business workflows is most of what we do, so we are specific about when it belongs. Before we recommend an AI feature, we compare it with what you do today and with the plain alternatives, such as a script, a rule, or a better form. If one of those does the job, we build that instead.
We do not add AI to a product so it can say it has AI.
02People stay accountable
We build with AI tools ourselves. A senior engineer reads and tests everything AI helped produce before it reaches you, and Manawaroa answers for that work exactly as it answers for work typed by hand.
In the AI features we build for you, your statement of work (SOW) sets which decisions the AI may make on its own and which a person approves first. You can always switch an AI feature off.
03Your code and your data
We may use AI tools on the code we write and maintain for you.
Your business data is different. Customer records, documents, financial information, and anything else your business holds go into an AI tool only when your SOW approves it, and only for the purposes the SOW lists.
04No training by AI providers
We only use AI tools whose terms do not allow the provider to train its models on what we send. That covers your code, your data, and our own work. If a tool's terms change to allow training, we stop sending it anything.
05Our own models
We want to get better at this work, and part of that is learning from the projects we do. If you opt in through your SOW, we may train our own internal models on the records of your engagement, such as the code we wrote, our notes, and how we solved problems. If you don't opt in, we don't.
We never use your business data or anyone's personal information for that training. Our models stay inside Manawaroa. We do not sell them, license them, or offer them to anyone else, and we do not use them to hand your code or confidential information to another client.
06What we will not build
We decline AI work designed to:
- Discriminate against people unlawfully
- Deceive people about whether they are dealing with a machine
- Watch people without their knowledge where the law requires consent or notice
- Break the privacy or confidentiality promises your business has made to others
07Changes to this statement
AI tools and the law around them change quickly, and this statement will change with them. We post the new version here with a new effective date. A change does not reduce the protections in an SOW you have already signed.
08Questions
Email responsible-ai@manawaroa.io with questions about how we use AI.