Straight answers about AI automation

New Zealand businesses of roughly 5 to 50 staff with a process that eats hours every week: quoting, invoicing, paperwork, scheduling, customer questions. I also take on larger organisations where the fit is right; I have delivered solution design and builds for a university and an investment firm. Not a fit: personal projects, courses, or learning to build AI yourself.

The useful test: if a task is repetitive and you can describe the rules for doing it, it is probably automatable. That covers reading documents and emails, moving data between systems, drafting quotes and replies, answering common questions, and chasing things that are late. The free call is where we work out which of your processes pass that test and what each one is worth.

Describe it. The capability list on the homepage is where most jobs land, not where they stop. Over twelve years I have built everything from document pipelines and company-wide AI assistants to full products with payments and accounts. If it can be specified, I can almost certainly build it, and if I think it is a bad idea I will say so.

No, it is the normal starting point. Most owners arrive with a feeling that something is slow or manual, not a spec. On the free call I ask about your week: where the double-handling is, what gets chased, what you dread on a Monday. From that we usually find two or three candidates, and I tell you which one to do first and why.

Trades and services, legal, retail and e-commerce, finance and insurance, property, health and allied health, education, and primary industry. The process matters more than the sector: quoting, intake, claims, scheduling and reporting look similar everywhere. Recent work includes a law firm, an air-conditioning company, a music company, a university and an investment firm.

Yes. Farms, packhouses, contractors and manufacturers tend to have the most paperwork per staff member in the country: compliance records, job sheets, supplier invoices, health and safety logs. Reading those automatically and pushing them into the systems you already use is exactly the kind of work that pays back fastest.

No. You need one person who knows the process and can answer questions, usually the owner or the office manager. I handle the technical side end to end and train whoever will use the system before it goes live.

Yes. The 30-minute call costs nothing and comes with no obligation. You walk me through the process, I tell you honestly whether automation will help, and if it makes sense you get a fixed-price proposal in writing. If it doesn't, you get a straight answer and no invoice.

Less than most people expect. Automation across the tools you already run usually lands between NZ$500 and NZ$5,000 and can be live in under a week. Document and email intelligence runs NZ$1,000 to NZ$8,000, a website NZ$1,000 to NZ$6,000, an AI assistant or company brain NZ$3,000 to NZ$12,000, and a custom app or portal NZ$4,000 to NZ$15,000. Every job is quoted as a fixed price after the free call, so your number is based on your process, not a menu. Small jobs, fixes and advice are billed hourly at a rate agreed upfront, and hosting and support once live run on a monthly plan from $99.

Most projects land between two and six weeks from proposal sign-off to live. Larger systems with several integrations can run two to three months, and I say so up front rather than discover it halfway through.

A good one costs $150,000 to $250,000 a year and will spend most of that time waiting for work. Most businesses under 50 staff need a few weeks of senior build time a year, plus someone on call when things change. That is what I offer, and if you ever grow to the point where a hire makes sense I will tell you and help you recruit them.

The free call can usually happen within a week. Builds typically start within two to three weeks of you approving the proposal.

Almost always. Where a system has an API, I connect to it directly: Xero, MYOB, simPRO, HubSpot, Salesforce, Shopify, Microsoft 365, Google Workspace, Slack, Zendesk, Akahu open banking and custom in-house systems. Where there is no API, UI automation can drive the software the way a person would.

By default on AWS in Sydney, the closest fully managed cloud region to New Zealand, inside an account you control. Where a contract or regulator requires it, I can build NZ-only or on-premises deployments. Nothing goes into training a public model.

Every assistant I build answers from your own documents and data, retrieved at the moment of the question, rather than from what the model remembers. It is told to say so and hand over to a person when it cannot find an answer, and every conversation is logged so you can see exactly what it said and why.

Everything runs in your own AWS account with access limited to the people and systems that need it, secrets kept in a managed vault, and every action logged. Data is encrypted in transit and at rest. Where you have specific requirements, such as NZ Privacy Act obligations, client confidentiality rules or an industry regulator, they go into the proposal and are built in, not bolted on.

Yes. The systems I build run on the same cloud services the banks use, so a thousand documents a day is not different from ten, just cheaper per document. The practical limit is usually the systems on your side, and I design around those.

Yes. An assistant, a quoting tool, a document upload that reads itself, or search that actually understands the question can all be added to what you already have, whether it is WordPress, Shopify, a custom site or a mobile app, without rebuilding it.

The AI models themselves need an internet connection, so no. What I can do is design for the failure case: forms and uploads queue locally and sync when the connection is back, and nothing is lost if a rural site drops out for an hour.

Zapier and Make handle fixed if-this-then-that triggers between apps, and they are the right tool for that. AI automation adds judgement: reading an email or a PDF, working out what it means, deciding what to do, and producing structured output, with a person checking anything uncertain. I use the simple tools where they are enough and only add AI where the input is messy.

Every system I run has monitoring and alerts, so I usually know before you do. Anything customer-facing I look at the same day during New Zealand business hours, and I tell you plainly what happened and what changed. I do not promise 24/7 support, because one person cannot honestly offer it.

You do. The code, the data and, wherever possible, the cloud account are yours. If you ever want to move it to another provider, everything is documented so you can.

Yes. The proposal fixes the scope and price, so a change is written down with its effect on cost and timing before it is built. Small refinements happen as we go; anything that shifts the shape of the project gets a short change note you approve.

If you have an internal developer or IT person, yes. Everything I build is documented, and I hand over with a walkthrough so your team can maintain and extend it. Some clients keep me on a support plan, some take it fully in-house after launch, and both are fine.

Yes, if you want me to. Hosting, monitoring and support run on a monthly plan from $99, and I stay on to keep the system working as your business and the tools around it change. There is no lock-in: the code and the account are yours.

A few hours a week of senior technical leadership without a full-time hire: roadmap, vendor decisions, security, hiring, and an honest read on where AI will and will not pay off. It suits businesses that have outgrown ad-hoc IT but are not ready for a $250,000 salary. It is a retainer, scoped to what you actually need.