Our work
Problems we've solved for clients, and things we've built for ourselves.
A recent engagement · an engineering business
Their experts built the first version.
- The problem
- The team knew exactly what they needed, but every requirements document lost something in translation.
- What we did
- Their experts built a rough working version with AI. We made it solid, tested it and put it into production.
- Where it is
- In use, and it fits the work, because the people who do the work designed it.
Your people know the work better than any document can.
More stories
-
01
Engineering software from the 1970s, rebuilt.
Result: three decades-old calculation programs turned into interactive tools the engineers can use and check.
An engineering consultancy whose design calculations still ran on Fortran written in the 1960s to 80s. Rebuilt in modern code, the know-how in them is no longer locked in software nobody can maintain.
The know-how was never the problem. It was locked in code nobody could read.
-
02
Checking the engineering, not just the wording.
Result: reports checked before they go out, with what's wrong and how to fix it.
The same consultancy. Reading through their project files, AI also found calculation tools giving results that weren't conservative enough.
Once AI can read your files, it finds problems nobody was looking for.
-
03
Finding buyers before the tender goes out.
Result: a ranked list of who's about to buy, with the evidence, run by their own team.
A Christchurch technology exporter selling to more than two thousand overseas agencies.
Their market wasn't the problem. Timing was.
-
04
A whole agency working with AI.
Result: Claude rolled out across the business, and staff building tools for their own jobs.
A Christchurch agency group. It started with talking to the people doing the work. Then came training, a company-wide AI competition, and tools the team built themselves.
Find the opportunity in your people. They already know where the work sticks.
-
05
Market research on tap.
Result: a research hub in their own Notion, filled by ChatGPT on a schedule, with the source behind every finding.
A New Zealand manufacturer's marketing team, who needed quick snapshots of their market, not expensive studies. We built it with them in a single hour-long session, in tools they already had.
You may not need new software. The tools you already pay for can do far more.
-
06
Who to follow up today.
Result: a short list each day of who to contact and why, kept in step with the CRM. Hours of research and writing saved.
The owner of a small creative studio, short on time for sales. Built in one session, in Claude. It reads his network, suggests who's worth contacting and drafts the message, all on one simple screen.
Small and specific beats big and general: one person, one job, one screen.
-
07
A memory for the business that stays current.
Result: meetings, emails and decisions filed where people and AI can find them, with the source behind each one.
We use it to run our own business and our clients' projects. Drop a document in, and AI reads it, files what matters and keeps the open questions in view.
AI is only as useful as what it knows about your business. Keep that current, and people trust it.
-
08
A radio station with nobody in the studio.
Result: a station that runs 24 hours a day, where a person can override any item before it airs.
A live system, not a demo. Built by our director Andy Masters.
AI can run on its own when a person can still say no.
-
09
A note-taker that keeps the meeting on track.
In development: everything a note-taker does, plus help while the meeting is happening.
Set an agenda and it shows, live, whether you're sticking to it. Decisions and actions are captured and assigned on the spot, with a one-tap OK. Built by Andy Masters.