Worth a 30-minute conversation?
The repetitive parts of your week, handed to software. I'm Balu, Kove is just me, and I build the thing myself rather than selling you someone else's.
Most of what saves a Canterbury business real hours has no artificial intelligence in it at all. It's plain automation: information moving between systems without anyone retyping it, approvals that chase themselves, documents that generate on their own, reports that are ready before the Monday meeting.
AI earns its place in the parts that need judgement on messy input. Reading an emailed invoice that's laid out differently every time. Sorting a support queue by what people meant rather than what they typed. Drafting a reply that a human then checks. Those are real jobs and they're worth doing well.
The distinction matters because one of them is much cheaper than the other, and plenty of people will happily sell you the expensive one for a job that didn't need it. I'll tell you which one you're looking at before you spend anything.
Quotes retyped into invoices, invoices retyped into Xero, job details retyped into a spreadsheet. The most common thing I fix, and usually the fastest to pay for itself.
Unpaid invoices, unsigned quotes, unreturned forms. Software is better at politely asking the same question every week than any person is, and it never forgets.
When off-the-shelf software fits about 70% of how you work and the other 30% is where your margin lives. I build the tool around your business instead of the other way round.
Power Apps, automated flows and Copilot agents on the Microsoft stack you're already buying. Often no new subscription at all. More on that here.
Agents that read messy input, sort it, and hand the hard cases to a human, a long way from a chatbot on your website. See what that looks like in practice.
An automation that breaks monthly. An app one person understands. I audit it, fix what's worth fixing, rebuild the rest, and document it so it never hangs on one person again.
If the process changes every time you run it, automating it makes things worse. If nobody in the business agrees what the process is, that's the problem, and no software fixes it. And under ten minutes a month, the maths doesn't work, so I'll say so.
I'd rather lose a job than sell you one that won't pay for itself. That's easier for me than most, because Kove is one person with low overheads and no sales targets to hit.
The fastest way to understand how work really happens is to watch someone do it for an hour. That's a short drive in Canterbury and a flight anywhere else.
The gap between how a process is described in a meeting and how it runs at 4pm on a Friday is where most automation projects fail. Sitting next to someone closes that gap in a morning.
Outside Canterbury I work over video and it works fine. It just takes longer to get to the same understanding.
A free 30-minute call, where I ask how your week runs and tell you plainly whether there's anything worth automating. No prep needed and no obligation. If there's something there, the next step is either a Workflow Review at $3,500 + GST that ends in a written plan you keep whoever builds it, or straight into a build. The full ladder is on the services page.
No. I'm based in Christchurch and I work with businesses across New Zealand. Being local just means I can sit in your office for a day when that's the fastest way to understand how the work actually runs, which for Canterbury clients it usually is.
Straightforward single automations typically land between $4,000 and $8,000 + GST. Bigger builds like a custom CRM or job-tracking system run $10,000 to $22,000. If you want an exact number before committing to a build, the Workflow Review is $3,500 + GST and ends in a written plan you keep either way.
If someone in your business retypes the same information into a second system, chases the same people every week, or builds the same report by hand every month, you're ready. If your process changes every time you run it, or nobody agrees what the process is, you're not. Fix that first, and automating gets much cheaper.
Usually not. Most of what saves a Canterbury business real hours is plain automation with no language model anywhere near it: moving data between systems, chasing approvals, generating documents. AI earns its place when a job needs judgement on messy input, like reading an emailed invoice or triaging a support queue. I'll tell you which one your job is, and the answer is often the cheaper one.
Yes, and for anything involving how a team actually works I'd rather. Watching someone do the job for an hour tells me more than a week of emails. Outside Canterbury it's video and async, which works fine, it's just slower to get to the same understanding.
You keep everything. Source code in your GitHub, credentials in your accounts, hosting on your bill, and documentation written for whoever comes next. Nothing I build depends on me still being around, because a one-person supplier that leaves you stranded is a bad deal and I'd rather not sell one.