The average New Zealand mortgage broker spends between four and six hours on the paperwork of a single deal. Application forms, document chases, file notes, lender follow-ups, refix instructions, compliance summaries, customer emails — the work is mechanical, but it adds up to most of the working week.
That ratio — hours of admin for every hour with a client — is the problem we started Twio to solve.
What we noticed
For the last decade, lender systems have improved at the margins, CRM tools have multiplied, and document workflows have moved from paper to PDF. But none of these changes addressed the structural issue: the broker is still the manual integration layer between the customer, the lender, and the paperwork.
The good brokers we talked to weren't drowning in any single task. They were drowning in context-switching. Twenty cases at different stages, each with its own email thread, its own documents, its own next action. Holding all of that in your head — while also being present for the conversation in front of you — is a kind of expertise that simply doesn't scale past a few dozen clients.
We thought the answer wasn't more software dashboards. It was an AI that could hold all of that context for you.
What we built
Twio is a workspace where the AI does the structural work — reading email, extracting customer information, filling forms, drafting proposals — and the broker does the work that requires judgement.
Three principles shaped what we made:
- The broker is always in the loop. Nothing leaves your inbox, your lender portal, or your client without you approving it. The AI is a research-and-draft partner, not an autonomous agent.
- Email is the central artefact. Almost every detail you need is already in your inbox. Reading email well — not asking you to re-enter everything into a CRM — is the highest-leverage thing the product can do.
- The model should disappear into the workflow. You shouldn't have to "prompt" Twio. The product knows what stage a case is at, what's missing, and what to do next. The AI surfaces through the structure of the work.
What this looks like in practice: you open Twio in the morning, see five cases that need your attention, and each one has a specific draft, decision, or follow-up ready for you to approve or refine. The work that used to fill your day is collapsed into review.
Why now
Two things changed in the last 18 months that made this product feasible.
The first is the obvious one — LLMs got good enough to read a customer email accurately, extract structured information, and write a coherent draft in your voice. Eighteen months ago, the error rate made this kind of automation impractical. Today, with the right scaffolding, it's reliable.
The second is the regulatory shift. The Financial Markets Authority published its guidance on AI use by advisers in 2024, with explicit expectations that AI tools augment — not replace — the adviser's judgement. That's exactly the model Twio is built around. We don't need to argue that AI should be in this workflow; the regulator already said so.
Who we're for
We've been deliberate about who Twio is for, and who it isn't.
Twio is for licensed New Zealand mortgage brokers and FAPs serving consumer lending. Our case templates, lender panels, and email patterns are tuned for the New Zealand market. We've built specifically around the FMCA 2013 framework and the AML/CFT obligations of broking businesses.
We're not aimed at general financial planners, investment advisers, or commercial finance specialists. There's nothing stopping you from using Twio if you do that work, but the product won't be uniquely good at it the way it is for mortgages.
What's next
We're in early access. The product is in use with a small group of brokers across New Zealand, and we're shipping improvements weekly. If you'd like to see Twio for your business, get in touch. We'll book you a 30-minute call, walk you through it, and — if it makes sense — get your tenant set up the same week.
Thanks for reading. We're glad to be building this in the open.