You rarely need to rebuild to add AI. Putti, an Auckland team that builds AI agents and integrations for New Zealand businesses, connects AI to the software you already run through APIs, grounds it in your own data and rolls it out one feature at a time.
Do you need a rebuild to add AI?
Usually not. Modern AI is delivered as a service your existing app can call. The work is integration and data plumbing, not starting over, so the risk and cost are far lower than a rewrite.
If your systems are Xero, MYOB, a CRM or an ERP, the same applies. Our post on adding AI to Xero, MYOB, your CRM or ERP goes through how, without replacing any of them.
How does AI connect to your system?
Three ways, and most projects use more than one. Through an API, your app sends a request to an AI model and uses the response, the way it calls any other service. Through your own data, using retrieval (RAG), the model looks up relevant records or documents first, so answers are grounded in your business and not generic. And through automation, AI triggers actions in tools you already use, such as updating a record or flagging a case for review.
How should you roll it out?
One feature at a time. Ship a single high-value feature first, such as an assistant, smarter search or document extraction. Keep a person in the loop, then expand once it earns trust. That's safer and cheaper than an all-at-once launch.
A proof of concept in 2 to 4 weeks is a quick way to test that first feature on your real data.
What's the Putti view?
We do AI integration that plugs into the systems you have, and we're upfront about what AI can and can't do reliably. If you're worried about what could go wrong, read how to stop an AI agent making costly mistakes. Or tell us what you want to add.
Last updated: November 2, 2026
Frequently asked questions
-
Can I add AI to my existing software without rebuilding it?
Usually, yes. Modern AI is delivered as a service your app can call, so the work is integration and data plumbing rather than a rewrite. Putti connects AI to the systems you already run, including Xero, MYOB, CRMs and ERPs. It's lower risk and lower cost than starting again, and you keep what already works.
-
How does AI connect to an existing system?
Three ways, often combined. Your app calls an AI model through an API, the way it calls any other service. The model retrieves records from your own data first, so answers reflect your business and not generic text. And AI can trigger actions in tools you already use, such as updating a record or flagging a case.
-
What is RAG and why does it matter for adding AI?
RAG, or retrieval-augmented generation, means the AI looks up relevant documents or records from your systems before it answers. That grounds the response in your actual data, which cuts the made-up answers a model can give on its own. For most business uses, it's the difference between a generic chatbot and something your staff can trust.
-
What's the best first AI feature to add?
The one with the clearest payback. Common starting points are a support assistant that drafts replies, smarter search across your documents, and extraction of data from invoices or forms. Pick a single high-value feature, keep a person in the loop, and expand once it earns trust. That's cheaper and safer than launching everything at once.
-
Is it safe to connect AI to my business data?
It can be, if it's designed carefully. That means giving the AI access to only the records its job needs, keeping a human checkpoint on anything that sends, pays or deletes, and logging every action. Our guide on stopping an AI agent making costly mistakes covers the controls in detail.
-
How long does it take to add AI to an existing system?
A focused first feature can be tested in 2 to 4 weeks as a proof of concept, then hardened for production. The timeline depends on how accessible your data is and how many systems it has to connect to. A short scoping conversation tells you whether it's weeks or months, and what has to be tidied first.