Case Studies

Systems we've built for Australian service businesses. Client names are left out; what we built and what changed are not.

A Melbourne migration firm

From hand-filled visa forms to a case system

The problem

Consultants re-typed client details into government forms such as Form 80, field by field, often translating from Chinese as they went. Client information lived in spreadsheets and chat threads, and it was hard to see where each case was up to.

What we built

  • A private intake link for each client: they fill in their own details, in Chinese if they prefer
  • AI translation of Chinese answers into English, and passport photo recognition that fills in name, date of birth and passport number
  • Automatic filling of Form 80 (734 fields) and Form 956, ready for staff to review
  • A staff dashboard with a 9-stage case pipeline, clients, leads, referrals and invoices, plus a mobile view
  • A bilingual Chinese and English website with a blog and visa guides

The result

  • 700–800 hours saved per consultant per year
  • Clients enter their details once; staff check the form instead of typing it

A non-bank lender

Rate updates from PDF to live website, with a human sign-off

The problem

Funders send new rate sheets as PDFs. Every change meant someone reading the PDF, updating product pages by hand and hoping nothing was missed. Brokers also had to run serviceability numbers in a spreadsheet.

What we built

  • AI that reads each rate sheet and extracts the new rates
  • A review screen showing what changed; nothing goes live until staff confirm
  • A bilingual website with product pages and rate tables that update automatically
  • An online serviceability calculator rebuilt from the broker spreadsheet
  • A printable product guide generated as a PDF, and AI-drafted client emails

The result

  • Rate changes go from PDF to the live site without re-typing
  • Staff approve every change before customers see it

An asset finance aggregator

Working prototype

An AI assistant that answers from lender policy, and shows its source

The problem

Brokers on the panel deal with dozens of lenders. Finding the right policy clause or rate meant digging through product guides, rate cards and credit policies one PDF at a time.

What we built

  • A document library of 200+ lender documents from 60+ lenders, including image-only rate cards read with OCR
  • Search that combines keywords and meaning, so brokers can ask in plain English
  • Every answer cites the lender, document and page; if it's not in the documents, it says so instead of guessing
  • A test set of real broker questions to check answer quality before launch
  • A marketing website with product, lender panel and industry pages

The result

  • Answers from 200+ lender documents, with the source cited

A law firm and an accounting practice

Bilingual websites for professional firms

The problem

Both firms serve Chinese- and English-speaking clients, and needed a website that explains their services properly in both languages — not a machine-translated copy of one.

What we built

  • Separate English and Chinese versions of every page, each indexable by Google
  • Practice-area and service pages written for the questions clients actually ask
  • Team profiles, articles and structured data so search engines and AI assistants understand the firm

The result

  • One site that works for both Chinese- and English-speaking clients

Fordexa (our own phone line)

An AI receptionist on our own phone

Every call to our number is answered by the AI receptionist we sell. It takes the caller's details, emails us a summary, and transfers to us during business hours. The easiest case study to check: call it.

Got a process that eats your team's week?

Tell us what it is. We'll tell you straight whether it's worth automating.