CASE STUDIES

The work behind the claims.

Two kinds of work, kept separate. Automation engagements delivered for clients, and platforms AUXA has built in-house. Below is what we built, how it works, and what went wrong on the way.

Platforms built

Software AUXA has built in-house. These pages describe what each platform does and how it is architected. They report no results, because none are published.

Automation engagements

Client work. The clients are not named because written permission to name them has not been given.

DELIVERED WORK

What has AUXA actually built?

Two engagements with the same New Zealand consultancy. The first put a dead website back online. The second turned that firm's consulting expertise into working software. Both are described in full below, including what went wrong on the way.

Custom SaaS

Turning an HR consultancy’s expertise into a compliance platform

A New Zealand HR and health & safety consultancy

What was the problem?

The consultancy delivered employment-relations and health & safety compliance as hands-on advisory work. Every client was served manually, so capacity was capped by consultant hours and the same policy work was rebuilt from scratch for each employer. They wanted their expertise to run as a product without losing the accuracy that made it worth paying for.

What did we build?

  • A multi-tenant web application where each employer organisation is isolated at the database level using PostgreSQL row-level security, not application-layer filtering.
  • A two-jurisdiction compliance engine covering both New Zealand and Australian employment law, with Australian state-level variation handled per employee rather than per company.
  • A statutory figures register in which every legal value carries its source and the date it was last reviewed, so no number in the product is unattributed.
  • Automatic obligation alerts driven from that register, including an annual re-indexation countdown for both jurisdictions.
  • Mandatory two-factor authentication for administrator accounts, enforced at the point of sign-in rather than left to an organisation setting.
  • A Xero integration for payroll and accounting data.
  • Continuous deployment: an automated test suite, including a full browser suite, gates every release to production.

What was the hardest part?

Adding a second country. Jurisdiction turned out not to be a property of the company — a New Zealand business can employ Australians. Making it a property of each employment instead of each organisation meant every rule in the system had to resolve whose law applied before it could answer anything. Six separate defects all had exactly that shape, so we wrote a test that fails the build whenever new code applies New Zealand law without first resolving jurisdiction. The class of bug cannot come back.

StackNext.jsPostgreSQLRow-level securityXero APIPlaywrightFly.io
Talk through a build like this
Recovery & rebuild

Recovering a business website that had been taken offline

The same New Zealand consultancy, before the platform build

What was the problem?

The consultancy’s website was down. It had been built by a previous developer on a no-code hosting platform, and that deployment had been broken and could not be restored. Enquiries were being lost every day it stayed offline. All that survived was a salvaged copy of the source; the content and leads held on the hosting platform were not in it and were at risk of being purged with the account.

What did we build?

  • A full rebuild on Next.js, chosen so the client owns the code and can never be locked out of their own site by a platform again.
  • Twenty-one pages restored and live: home, nine individual service pages, about, pricing, insights, contact, and the legal pages.
  • Working lead capture with a server-side fallback that logs every submission, so no enquiry is lost even if the email channel fails.
  • Structured data, an AI-crawler-friendly robots policy, and an llms.txt summary so the site is legible to answer engines as well as to Google.
  • A self-removing script to clear the stale service worker the old platform had installed, which was still serving the dead site to returning visitors from their own browser cache.

What went wrong, and what we changed because of it?

The contact form passed every test and delivered nothing. The email provider returned HTTP 200 on failure, with the actual result buried in the response body — and it silently rejected server-to-server calls from cloud hosting IPs, which is exactly how our server was calling it. Every green test had been a false positive. We only found it by checking whether a real email arrived rather than whether the request succeeded. The fix moved submission to the browser, where the provider accepts it. We now test delivery, not status codes.

StackNext.jsTypeScriptTailwind CSSVercelJSON-LDllms.txt
Talk through a build like this
QUESTIONS

Why is there no client name on this page?

The honest answers to the questions this page raises.

Why do these case studies not name the client?

Because we do not have written permission to. AUXA does not publish a client’s name, logo or results until the client has agreed in writing to the exact wording. Everything described here is work we actually did and can walk you through in detail on a call — we just will not put someone else’s name on a marketing page without asking them first. You would want the same.

Why are there no percentages or hours-saved figures?

Because we have not measured them. Publishing an unmeasured improvement figure is a false representation under the Fair Trading Act 1986, and it is also the fastest way to lose a technical buyer. Where we quote a number on this site, it is one we can evidence. Where we have not measured something, we say nothing rather than estimate.

Can I speak to a reference before signing anything?

Yes. Reference calls are arranged directly with the client rather than published, which is the same protection you would get. Ask on the audit call and we will introduce you.

What size of engagement does AUXA take on?

Both ends. A focused single-workflow automation is a small, defined piece of work with a fixed starting price. A multi-tenant SaaS platform like the one above is a continuing build with releases shipping to production on an automated pipeline. The audit call establishes which of those your problem actually is.

CAPABILITIES

What We Build

The kinds of automation and digital systems we build: lead and CRM workflows, reporting stacks, and internal operations tooling.

Lead & CRM

CRM Lead Response System

GoHighLevel and n8n pipeline that routes new leads, drafts follow-ups, updates deal stages, and alerts the sales team when urgency is high.

GoHighLeveln8nOpenAI
Discuss a similar system
Reporting

Marketing Reporting Stack

Campaign data, store revenue, content tasks, and weekly summaries connected into one operating view for faster budget decisions.

Meta AdsClaude AIMake.com
Discuss a similar system
Internal Ops

Notion Knowledge OS

Client delivery, SOPs, project status, and recurring tasks organised in a Notion workspace with n8n reminders and management reporting.

Notionn8nAI SEO
Discuss a similar system
GET IN TOUCH

Ready to Transform Your Business?

Book a focused automation audit. We will identify the workflow most likely to save time, increase speed, or improve lead conversion first.

Enquiries
Use the audit form — it reaches us directly
Phone / WhatsApp
+64 21 465 109
Coverage
New Zealand and Australia — remote delivery
Response Time
Within 24 hours · Same-day usually

What Happens Next?

1

We review your current process, tools, and obvious revenue leaks.

2

We map the highest-ROI workflow to automate first, with risk and effort noted.

3

You get a clear roadmap: what to build, what to skip, and what it should return.

Get Your Free Automation Audit

No spam. No pressure. If automation is not worth it, we will say so.