To start using AI, pick one repetitive, high-volume workflow, prove the value on a small pilot, measure the result and then expand. Putti, an Auckland team that builds AI agents for New Zealand businesses, sees the fastest wins in support, document processing and data entry. Here's a practical way in.
Where should you start?
With the problem, not the tool. The mistake most businesses make is saying "we should use AI" before naming a problem worth solving.
Look for work that's repetitive, high-volume and follows patterns a person currently handles by hand. Those are the tasks where AI pays back quickly. If you can't name the workflow and the hours it costs you, it's too early to pick a tool.
Where does AI usually pay off first?
Customer support is a common first win, where AI drafts replies, answers common questions and routes tickets. Document processing is another: reading invoices, forms and contracts and extracting the data. Data entry and cleanup comes next, moving information between systems accurately. And search lets staff ask questions of your own documents.
In each case the AI handles the routine and your team handles the exceptions. Our post on how AI agents cut admin work in a NZ business shows what that looks like job by job.
How do you run a small pilot?
Pick one workflow, set a clear success measure such as time saved, error rate or response speed, and run a contained pilot before committing. Keep a human in the loop on anything customer-facing, and expand only once the numbers hold up.
A proof of concept in 2 to 4 weeks is how we usually run that first pilot, and what AI implementation costs in New Zealand breaks down the money.
What's the Putti view?
AI is a tool, not a magic wand. We build custom AI solutions around real workflows, and we're always upfront about where AI helps and where human judgement stays in charge. If you'd rather talk it through first, our AI consulting starts with your workflows, not a product pitch. Or tell us your workflow.
Last updated: November 2, 2026
Frequently asked questions
-
How can a small business start using AI?
Start with a problem, not a tool. Pick one repetitive, high-volume workflow, such as replying to common customer questions or reading invoices. Set a clear success measure, run a small pilot, and expand only when the numbers hold up. You don't need a data team or a big budget to begin, just a well-scoped first task.
-
Where does AI pay off first in a business?
Usually in customer support, document processing and data entry. AI drafts replies, reads invoices, forms and contracts and extracts the data, and moves information between systems accurately. It handles the routine and your team handles the exceptions. Search across your own documents is another common early win.
-
How do I run an AI pilot?
Pick one workflow and one success measure, such as hours saved, error rate or response time. Run it on real examples from your business, including the awkward ones. Keep a person checking anything customer-facing. If the numbers hold after a few weeks, expand. If they don't, you've learned that cheaply.
-
How long does it take to see value from AI?
A focused pilot can show results in a few weeks. Putti builds proofs of concept in 2 to 4 weeks, so you can see whether an idea works on your real data before you fund a full build. Value then grows as you extend the working feature to more of the workflow.
-
How much does it cost to start using AI in a NZ business?
It depends on scope, but a first pilot doesn't have to be large. A proof of concept runs between $5,000 and $15,000 NZD plus GST in most cases, which is a fraction of a production build. Our post on what AI implementation costs in New Zealand breaks down the drivers in detail.
-
Do I need to hire an AI specialist to get started?
No. You need someone who owns the workflow and can say what good looks like, plus a team that can build and integrate it. Putti handles the building and the connection to your systems. Your part is describing the work honestly, including the messy parts, because that's what the AI will actually run on.