Systems integration
Your online store, ERP or back-office, accounting and marketing platforms connected so data flows automatically — instead of being re-keyed by hand.
[ Independent IT consultancy · Australia ]
One technical partner for the whole stack.
From the database to the storefront to the server room — I connect the systems your business runs on, so they finally work as one.
Trading for over a decade. Small by choice — you deal directly with the person doing the work: someone who has run businesses, not just built software for them.
[ 01 · The gaps ]
Most problems don't live inside one system.
They fall between them.
The e-commerce developer doesn't touch the ERP. The web agency doesn't do servers. The IT support company doesn't write code. And nobody owns the data flowing in between. That middle ground — the part everyone else leaves in the gaps — is exactly where I work.
-
“Our systems don't talk to each other.”
So staff re-key the same data three times. Integrated properly, it flows on its own.
-
“We waste hours on the same manual job every week.”
Automated, it just happens — on schedule, without the errors.
-
“Our data is a mess, and nobody trusts the reports.”
Cleaned up, moved to where it belongs, and reported from one source of truth.
-
“We need a tool that doesn't exist off the shelf.”
Then it gets built — around how your business actually works.
-
“We depend on an ageing system nobody's game to touch.”
Handled carefully: understood first, then modernised or connected — without breaking what works.
-
“Nobody really manages our IT. We just hope it keeps working.”
Servers, backups, networks and monitoring — quietly looked after, so it does.
If any of those sound like your business — a retailer, wholesaler, distributor or service business running real operations on systems that don't quite cooperate — you're exactly who this site is for. You don't need a big agency or a full-time hire. You need one capable person who can see the whole picture.
[ 02 · What I do ]
Outcomes first. Then the technology.
Every engagement starts with the problem, not the product. These are the areas I cover — and because it's one person covering all of them, the joins between them are covered too.
Automation, data & analytics
The weekly manual jobs — data pulls, reports, reconciliations, imports and exports — replaced with reliable pipelines that just run. Plus the analysis on top: forecasting, margins, sales reporting you can actually trust.
Custom software
Bespoke web and mobile apps, internal dashboards and process tools, built around a specific business problem when nothing off the shelf fits.
E-commerce & websites
Online stores and websites built, customised, migrated and hosted — including product data and catalogue work — kept fast and maintainable.
Cloud & infrastructure
Cloud hosting, servers, deployments, DNS, and migrations off legacy platforms — the modern plumbing your systems sit on.
IT operations & networking
Day-to-day systems administration, multi-site networking, monitoring and management, backups and disaster recovery.
Security & hardening
Practical security woven through all of it — access control, credential hygiene, backups that actually restore, and careful, reviewed changes.
Not sure where your problem fits?
That's normal — most real problems span two or three of these. Describe it in your own words and I'll tell you where I'd start.
Start a conversation[ 03 · Recent work ]
A few things built recently.
No client names here — just the shape of the work, recent and ongoing. Happy to talk through any of them in detail.
Competitor price intelligence
For a national buying group: a platform that keeps the product catalogue synced from their PIM and e-commerce systems, monitors competitor pricing across the market, and serves it all up as a live dashboard for pricing decisions.
Storefronts → marketing platform
A production sync feeding customer and order activity from a whole fleet of e-commerce sites into an email-marketing platform — so campaigns run on live data instead of stale exports.
Sales & P&L dashboard
For a consumer-brands wholesaler: a web dashboard of sales actuals versus forecast, P&L views and retailer-deduction analysis across brands and accounts — backed by a data warehouse and BI cubes. Recent additions: drill-down from P&L lines to the source trading-terms records, a risks-and-opportunities sheet moved out of a shared macro workbook into a live page several people can edit at once, self-service subscriptions to around twenty scheduled reports, and an order-fill tab that reproduces the supply team's weekly email to the cent.
An AI analyst for the numbers
A companion to that dashboard: an AI service that answers plain-English questions over sales, GL and forecast data — top line down to item level — by writing and running its own queries against the warehouse. Locked to read-only, every query audit-logged, and it asks a clarifying question rather than guessing.
Market share, every week
For a consumer-brands wholesaler: a weekly feed of syndicated supermarket scan data drops straight into the analytics dashboard, showing brand share against competitors, in-store versus online sales, and the new products and competitors worth watching. Everyone with a login gets an email when the new week lands.
Orders typed in plain English
Staff describe an order in their own words, or upload a customer's purchase-order PDF. An AI assistant fills in the order form, the system validates every line against the ERP, and a person confirms before anything posts. The AI only ever sees product and account codes, never customer names; PDFs are read on-premises, and every instruction is audit-logged. Now rolling out.
Supplier delivery-date chasing
For an industrial distributor: finds purchase orders with stale delivery dates in the ERP, emails suppliers a no-login link to confirm new ETAs, and writes approved dates back — replacing a manual chase that soaks up customer-service hours. In build.
A secure bridge into a legacy ERP
For the same distributor, whose ERP has no supported API or export: a small agent on the client's own server collects approved read-only queries from a cloud service over outbound HTTPS and sends the results back encrypted. No inbound port is opened and no raw SQL is ever accepted — every query comes from a fixed, versioned catalogue, each one proven against both of the client's companies. Security-reviewed and backed by more than 3,600 automated tests.
Subscription app, concept to launch
A client's new venture taken end-to-end: a subscription web and iOS app with AI-assisted features, payments and a production cloud backend — designed, built and shipped.
An inbox that answers itself
Rules-driven mailbox automation for a busy operations inbox: files attachments, parses incoming reports, and answers routine data questions by querying the ERP and data warehouse — unattended, around the clock.
Order reconciliation, storefront ↔ ERP
Daily jobs for a multi-brand wholesaler matching online-store orders against the ERP — auto-cancelling stale orders (with safety caps) and flagging any order that never made it across.
Online orders into the ERP, with nothing missed
For a consumer-brands wholesaler with five online stores: a replacement for an ageing file-drop integration. Every paid order becomes exactly one ERP sales order, despatches flow back to the storefront as tracked shipments, and stock levels sync out with a safety buffer. Each write is keyed and recorded before it's sent, so nothing posts twice, and every flow has its own independent back-check. The new stock calculation already matches the existing feed on all 465 products. Cutover will go store by store, with the new checks running alongside the old path first. In build.
Fleet-wide e-commerce changes
A command-line tool that applies one change specification across an entire fleet of e-commerce tenants — dry-run first, a canary site next, then the rest, with post-write verification and a full audit log. Recent rollouts include search-ranking boosts across 53 sites, a new brochure page across 52 and a product-grid change across 50. When a storefront outage turned out to be an expired security certificate rather than the supplier's system, a fleet-wide certificate check followed.
Sales & P&L forecasting engine
A forecasting system for a wholesale business that has grown into a finance platform: a calculation engine guarded by a regression suite, a claims desk with deal-linked write-offs, brand-level P&L reporting with cost allocation, and month-end packs that go straight to the accountant.
Retailer claims reconciliation
Pipelines that read the deduction claims major retailers send a supplier and reconcile them against the promotional-spend commitments recorded in the ERP — with a weekly exceptions email so nothing slips through.
School booklist platform
A web application for managing school booklists across a group of dealer websites, syncing lists and products into their e-commerce platform.
Reporting & document automation
A 7am email with month- and year-to-date sales, pending orders, stock on hand and overnight inventory movement, so the day's numbers are read before the day starts. Alongside it, invoice PDFs pulled from a mailbox, their key fields extracted, filed into a dated archive and routed to the right destination by rule.
Nightly ERP extracts
A modern replacement for ageing extract tooling: pulls data from a legacy ERP each night and reloads the reporting databases the business's dashboards and analysts read from.
Brand-site migrations
Several brand websites rebuilt as fast static sites and moved onto edge hosting — old URLs preserved so search rankings survive the move.
AI-assisted ticket triage
A tool that reads a support queue, has AI analyse each ticket and recommend an action, and keeps a human approving every one before anything executes.
Website performance forensics
A monitoring probe that watches site response times around the clock and automatically captures a full evidence pack the moment they spike — proof you can take to a hosting provider.
Month-end, without the spreadsheets
A rolling programme replacing more than a dozen fragile finance workbooks — freight-allocation journals for three carriers, standing journals, an annual regulatory ingredient declaration, monthly overdue-account letters — with tested Python pipelines. Each one is backtested against years of history before cutover; one caught three hand-typed errors the old process had quietly shipped.
Retailer data, on tap
Unattended pipelines that pull a supplier's data from two major retailers into one warehouse: lost-sales, scorecard and margin reports from a supermarket's supplier portal, and a pharmacy chain's weekly sales feed. The pharmacy feed replaced a legacy desktop database; two years of history were migrated and reconciled row by row, with zero discrepancies, before cutover.
Printer-cartridge finder
A storefront finder — pick the printer brand and model, see the cartridges that fit — that had been blocked for two years because supplier data carried no matching product codes. Cleared with a pipeline that derives the matches from OEM codes and compatibility data, and keeps the finder current as supplier files refresh. Now live on 48 dealer storefronts.
An order watchdog
For a trade storefront selling into two countries: a service that catches the worst failure in e-commerce — a customer's card is charged, the success page shows, and no order is ever created. It also confirms that every order's notification email was actually delivered, reconciling against the mail provider's log. One alert per incident plus a daily digest; detection only, no noise.
Two systems, no shared key
Nearly nine hundred retail stores in one system, debtor accounts in another, and nothing in common but store names and postal addresses. An evidence-scored fuzzy-matching engine paired them one-to-one and produced a confidence-ranked review workbook for human sign-off — instead of weeks of eyeballing.
Multi-region storefront estate
Ongoing engineering across one brand's four regional online stores — navigation and content features, analytics event tracking, cookie-consent work, and heavy third-party scripts tamed behind load-on-click facades verified to cost zero page speed.
A whole shop display, ordered in one click
A merchandising program on a trade storefront: the customer picks four display modules, sees exactly what is in each, and one click writes the entire stand to their order in a single request. A companion page handles restocking — each module expands into its component products with a quantity box per line. Module contents come straight from the ERP's kit records, so the storefront never holds a second copy of the product data. Launching soon.
Catalogues that follow the dealer
Digital flipbook catalogues published once and embedded across roughly sixty independent dealer storefronts. A script inside each catalogue rewrites every head-office product link to whichever storefront is embedding it, so a shopper who clicks a product in one dealer's catalogue lands on that dealer's site — not the corporate one.
A storefront that knows where you are
Geolocation for a storefront selling into two countries: a visitor resolving to New Zealand is offered the New Zealand site instead of bouncing off the wrong catalogue and pricing. A second mode, built and held behind a flag pending a business decision, hides pricing and add-to-cart from visitors outside the region and offers an enquiry route.
Pick the right gas strut from three measurements
A product selector for a hardware supplier: the shopper enters door weight, width and opening angle, and the page narrows a few hundred candidate struts down to the single best fit — then reports the frame mounting point in millimetres, so it can actually be installed. Anything outside the supported range hands off to a contact form with the measurements already carried across, so an unusual job becomes an enquiry rather than a dead end. Running across three storefronts in two countries.
Web analytics for fifty-odd storefronts
For a buying group whose members each run their own online store: individual monthly web-analytics reports for more than fifty members, all generated from one master template, with the monthly email schedules set up automatically and close to a hundred stale reports cleared out.
Menu boards straight from the till
Five menu boards whose prices lived in slide decks, separate from the point-of-sale system, and the two had drifted apart. Now staff export the catalogue from the till, choose the file in a small Windows program, and get all five boards back as finished images, with a list of every price that changed to check first. The hard part was the matching: the till held near-identical items at different prices, and in one case two entries with the same name and the same price. So each of the 119 prices is tied to one specific till item by its ID, not its name, and the shop signs off every pairing. If a single price can't be matched safely, the program stops and writes nothing rather than put a wrong price on the wall. The new prices are drawn to within three pixels of the original artwork, so a regenerated board looks like the one it replaces. Backed by more than 500 automated tests, with a one-page guide written for non-technical staff.
…and the quiet upkeep
Online-store themes, WordPress sites and portals, storefront widgets, redirect servers, Windows and Linux servers, backups and networks — the everyday care that keeps businesses running.
[ 04 · How I work ]
Careful with the systems your business runs on.
-
1
Understand the business first
Before any technology talk: what's actually costing you time, money or sleep? I learn how the work really happens — not how the org chart says it does.
-
2
Pick the pragmatic solution
The simplest thing that will hold up long-term — maintainable, honest about trade-offs. If something's a bad idea, I'll tell you before you spend money on it.
-
3
Deliver safely
Changes are tested, run against a small sample first where the risk is real, and shipped with a rollback path. Production systems get production-grade care.
-
4
Stick around
I work with businesses over years and many projects, learning how they operate as I go — an embedded technical partner, not a churn-and-burn contractor.
[ 05 · About ]
One person. The whole picture.
I'm Heath Wright — an independent Australian IT consultant, and the person you'll deal with from first email to finished work. The Wright Consultant has been trading for over a decade, and it's small by choice: no sales team, no bench of juniors, no hand-offs. And the client I started with in year one is still a client today.
The path here is unusual. Before consulting, I spent two decades on the commercial side of retail and FMCG — supermarket management, category analysis for a global beverage company, retail scan-data analytics, and national account management selling into the major Australian supermarkets. In every one of those roles I ended up building the systems too: automating a five-day business review down to one day, constructing forecasting models, programming the reporting databases. Forecasting, ranging, trade spend and P&Ls aren't abstractions to me — they've been my job.
It's still true today: alongside consulting, I run a wholesale distribution business as its general manager — order processing, AR and AP, demand forecasting, a remote warehouse, and the P&L. When I automate a reconciliation or build a forecast, it's because I've lived the manual version. I read a P&L as comfortably as a stack trace.
On the technical side, the work runs right across the stack — legacy business databases and ERPs, online stores, data pipelines, custom applications, cloud servers and networks — and, more importantly, the connections between them. That breadth is the point: I can carry a problem all the way from “this is painful” to “this is solved and running reliably.”
Based in Inverell, NSW. Most of the work happens remotely — wherever your systems are.
Full-spectrum, in one person
The database and the server and the app and the integration — and how they fit together.
Business-first, by experience
I don't just build for businesses — I run one. Forecasts, margins, cash flow and month-end are familiar ground.
Safe delivery
Tested changes, canary rollouts where risk is real, rollback paths. Your systems stay up.
Direct and honest
No jargon-fog, no over-promising. If something's a bad idea, you'll hear it from me first.
A long-term partner
Years-long relationships across many projects — my very first client is still a client today. Not a contractor who vanishes at handover.
Current
Modern cloud-native, automation-first tooling and practices — not stuck doing things the slow old way.
[ 06 · Under the hood ]
For the technically minded.
Everything above is deliberately plain English. This bit isn't. If you're the person who'll sanity-check the consultant, here's the actual footprint.
Languages
Python · SQL (SQL Server, PostgreSQL, Informix, Db2/AS400) · JavaScript / TypeScript · C# Razor · PHP · VBA · Liquid
Frameworks & apps
Django · FastAPI · Dash · Next.js / React · Celery · pytest
Data & BI
SQL Server + SSAS cubes (MDX) · PostgreSQL · Redis · SQLite · Power Query / PowerPivot · Phocas · forecasting engines · ETL pipelines
Commerce & CMS
Shopify · BigCommerce · WordPress (custom themes) · Commerce Vision
ERP & business systems
Pronto Xi (including its 4GL and Informix schema) · Infor M3 / Lawson on IBM iSeries · MYOB · Microsoft 365 / Graph API
Cloud & infrastructure
AWS · Cloudflare (Workers, DNS) · Docker · nginx · Linux (Fedora, Ubuntu) · Windows Server · KVM virtualisation · UniFi networking · encrypted Borg backups
AI & automation
Anthropic & OpenAI APIs · local models (Ollama) · RAG / vector search · MCP tooling · guard-railed SQL agents · RPA · scheduled pipelines
The long game
SQL off an AS/400 since 2011 · production VBA since 2013 · a ~250,000-line personal knowledge base reverse-engineering a proprietary ERP language · 2,900+ documented notes on one commerce platform's internals
Built by someone who knows how attackers work.
I'm not a pentest firm, and I won't pretend to be one. But I train the way attackers do, continuously: ranked around the top 500 of TryHackMe's ~2 million users at peak — Australia's monthly #1 in March 2024 — with ten completed learning paths across offensive tracks (Jr Penetration Tester, Offensive Pentesting, Red Teaming) and defensive ones (SOC Level 1, Cyber Defence), plus documented hands-on lab work across the full attack chain: reconnaissance, exploitation, privilege escalation.
What that looks like in your systems:
- Least-privilege by default — read-only guards and audit logs on anything that touches data. The AI warehouse analyst above is locked to SELECT-only.
- Credential hygiene — SSO over long-lived keys, secrets kept out of code, automated secret scanning in the development workflow.
- Segmented networks and isolated environments for sensitive work.
- Encrypted backups that get restore-tested — not just scheduled and hoped about.
- Careful change management — canary rollouts and rollback paths on the systems your business runs on.
- Current defensive tooling knowledge — SIEM/XDR (Wazuh, Security Onion), host intrusion detection, file-integrity monitoring, DFIR.
[ 07 · Contact ]
Tell me what's not working.
A short email is plenty. Describe the problem in your own words — you don't need to know what the fix is called. I'll tell you honestly whether I can help, and what I'd look at first.
- Based in Inverell NSW, Australia
- Reach me Use the form — it lands straight in my inbox.