AI and automation that actually ships.

The repetitive, rules-based work your team does every week: processing emails and documents, drafting quotes and replies, moving data between systems, chasing approvals, and answering the same internal questions. If someone does it the same way every time, it's usually a candidate. Two real examples from our client Waterware show the range. Their finance team was manually raising sales orders and reconciling purchase orders against invoices — we built an automation that reads the supplier PDFs and does both, cutting the re-keying and the entry errors that came with it. Their technical team had decades of product knowledge locked in a few heads, so we built Gandalf, an internal AI chatbot trained on their own knowledge base; now anyone can ask a plain-English question and get the answer, the part number, stock levels, and tracking from one interface. Both were weeks of work, not months.

Most engagements ship something useful within weeks, not months. We follow a four-step rhythm: scan your business for opportunities, assess which ones are worth doing, build the automation, then embed it with your team so it actually gets used. The first working automation is usually live inside the first month. Waterware's Gandalf chatbot is a fair benchmark for the pace: from concept to first live working model took four to five weeks, and that was a substantial build — an AI assistant trained on their whole product knowledge base, not a toy. Their director Darren Yley described the process as easy and the speed as amazing, which reflects a deliberate choice on our part: we'd rather prove value with a small working tool quickly than spend a quarter producing a strategy document. Each delivered automation also teaches your team what's possible, so the second project is usually better-chosen than the first.

Yes, if it's set up properly — and that setup is exactly where an IT and security background matters. We configure AI tools so your data stays inside your Microsoft 365 tenancy or the vendor agreements you already have, and it's never used to train public models. Access is scoped so an internal chatbot only sees what it should: a knowledge assistant like Waterware's Gandalf can read the product knowledge base without touching payroll or HR records. Because we're a managed IT and security provider first, the questions most AI vendors skip — where does the data live, who can query it, what happens when a staff member leaves, how is access logged — get answered as part of the build rather than discovered afterwards. We'll also tell you plainly when a tool's data terms aren't acceptable for New Zealand privacy obligations, and find one whose terms are.

No, and the tools we build are deliberately designed for teams without technical staff. We do the technical work — scoping, building, connecting to your systems, securing access — and hand over tools your team runs day to day in plain language. Waterware is the proof: a 35-person plumbing and heating importer, not a tech company, whose whole team now uses an AI chatbot daily. The design goal there was specifically that someone with very little product knowledge could ask a question in plain English and get a clear answer — no query language, no training course, no dependence on the one person who understands the system. Our workshops are pitched the same way: they're built for business owners and their staff, not developers. What we do need from you is the thing no consultant can supply — knowledge of how your business actually works, and honesty about where the time goes.

It depends on scope, which is why we start small and deliberately: a two-hour workshop or an opportunity scan comes first, and you get a prioritised list of automations with the expected time savings for each, so you can decide what's worth building before committing to anything. As a sense of scale, Waterware's internal AI chatbot — a substantial build, trained on their whole product knowledge base — took four to five weeks from concept to working model, and their director described the investment as clearly worthwhile relative to the key-man risk it removed. The honest way to think about cost is against the hours you're paying for now: a business that wins back a day a week of admin time has effectively hired a fifth of a person for the price of a small software build. If the numbers don't stack up for a given automation, the prioritised list shows that too, and we'll say so.

Book the two-hour AI workshop. We research your business beforehand, then work through your actual processes with your team in the room, and you leave with a shortlist of automation opportunities specific to your business, ranked by effort and payoff. From there you choose what to build first — with us or without us; the list is yours either way. If you'd rather start even more informally, get in touch and describe the most annoying repetitive job in your business; that conversation is usually the fastest way to find out whether AI can take it off your hands. And if you want to see what the destination looks like before starting, read the Waterware case studies — the Gandalf knowledge chatbot and the finance-ops automation both started exactly this way: one well-chosen problem, a few weeks of focused work, and a tool the whole team now relies on.