Custom software development in Australia
ComplxAI is a custom software development company in Australia. We build the software a business runs on: internal platforms, operations systems, customer and partner portals, job-management and workflow systems, integrations, data platforms and mobile apps. AI features go in where they earn their place, not as the headline.
- The IP is yours from the first paid stage
- Working software every couple of weeks, in 12-week blocks
- Documentation and handover included; take it in-house when you choose
Software your business runs itself
This page covers software built for one organisation: the system your operations team logs into each morning, the portal clients book through, the app your field crew carries. A subscription product sold to other businesses is a different job, covered on our SaaS development page.
If your internal tools are spreadsheets, shared inboxes and a product that fits almost well enough, the job is replacing that mix with one system built to the process. Where an existing product only needs new capability, AI feature development or an architecture review is the better first step.
A good fit if…
- Operations run on spreadsheets, email threads and a shared drive
- An off-the-shelf tool covers most of the process and staff work around the rest
- People re-key the same data into Xero, a CRM and a job system
- You want to own the software outright rather than rent it by the seat
Three reasons businesses commission custom software
The trigger is rarely software; it is a process that has outgrown the tools holding it together.
The spreadsheet that became the system
A workbook started as a tracker and is now the source of truth for jobs, stock or clients. Two people understand it and nobody can see who changed what. We rebuild it as an application with roles, history and validation.
Tools that fit most of the process
The off-the-shelf product handles the standard case; your exceptions live in workarounds and a second tool nobody planned for. We start from the process as it runs, so the edge cases are designed in.
Systems that do not talk to each other
Quotes in one place, jobs in another, invoices in Xero, contacts in HubSpot, and a person copying fields between them. We build the layer that keeps them consistent. Our AI integration services apply the same discipline when a model sits in the middle.
Six system types we build
Six shapes a brief can take, each with an example. A web application is the base for all six, on a shared API so a portal or mobile app can follow without a rebuild.
Internal platforms and operations systems
Jobs, clients, stock, rosters and compliance records in one place, with roles, audit history and screens designed with the people who use them daily.
In practice: a dozen job, staff and compliance workbooks become one application.
Customer and partner portals
Self-service for the people outside your business: bookings, document exchange, approvals, payments and status, with a clean line between client and staff views.
In practice: clients book, sign and pay against the same record staff see.
Workflow and job-management systems
Work moving through defined stages, from enquiry to quote to job to invoice, with a person approving wherever money or a commitment moves.
In practice: quotes priced from a catalogue; margin reconciled at job close.
Integrations and APIs
Connections to Xero, MYOB, HubSpot, Salesforce, email, calendars and document stores, with idempotent writes so a retry never duplicates an invoice, and a custom API when partners integrate with you.
In practice: HubSpot, Xero and the industry system stay, kept consistent by webhooks.
Data platforms and dashboards
Operational data pulled out of the systems that trap it, modelled once in Postgres or S3 and reported through dashboards an owner actually opens.
In practice: utilisation, margin and pipeline by person, job and client.
Mobile and field apps
Job sheets, photos, signatures and time on site from a phone, on the same API as the web application. Whether it must work out of range is settled in scoping, because that shapes the data model.
In practice: the closed job reaches Xero as an invoice, nothing re-keyed.
The stack behind our custom software
Mainstream, well-supported technology any competent team can maintain: serverless or containers as the workload demands, Postgres for anything relational, in your own AWS account or one we set up for you, all of it described as code. If the platform is the open question, our cloud engineering work covers setup, migration and cost.
- Interfaceswhat people useTypeScriptReactNext.jsMobile appsDesign systemCloudFront
- Servicesbusiness logicNode servicesPython servicesAPI GatewayLambdaECS/FargateStep FunctionsEventBridgeSQS
- Datastorage and reportingRDS/PostgresDynamoDBS3Postgres + pgvectorPoint-in-time recovery
- Integrationsthe systems you keepXeroMYOBHubSpotSalesforceEmail and calendarDocument storesWebhooksCustom B2B APIs
- AI featureswhere they earn their placeOpenAI, Anthropic, Google and Bedrock APIsDocument extractionRAG over your recordsEvaluation harnessApproval steps
- Foundationssecurity and operationsCognitoLeast-privilege IAMSecrets managementCloudWatchTerraform/CDKCI/CDStaging environments
Shape, build, ship, extend
A custom build follows the five-stage engagement described in how we work: two free conversations, then paid stages, each priced in writing before it starts.
Ownership and support
- The IP is yours from day one of any paid stage
- Code lives in your repositories and runs in your own AWS account, or one we set up for you
- Infrastructure definitions, tests and documentation are part of the handover
- A 14-day support window after launch, then paid support if you want it; no retainer required
- ShapeDiscovery and, where needed, Strategy: process mapping with your team, a use-case map scored by impact, effort and risk, and the build, buy or defer call. Data model, service boundaries, integration contracts and security model are recorded before code starts.
- Build in 12-week blocksWorking software every couple of weeks, starting with the workflow that hurts most. Unit and integration tests in CI, end-to-end tests on the critical paths, staging that mirrors production and a review before each release.
- Ship and hand overProduction release, migration from the old sheets and tools, architecture overview, runbooks and admin guides written as we go, and a handover session for whoever maintains it next.
- ExtendAdd the next workflow, a portal, a mobile app or an AI feature in a further block, with us or your own team.
Off-the-shelf, custom, or custom with AI
The answer depends on how standard the process is and what the workarounds already cost.
| Consideration | Off-the-shelf | Custom | Custom + AI |
|---|---|---|---|
| Fit to process | Fits the standard case; exceptions become workarounds. | Built to the process as it runs, exceptions included. | As custom, plus judgement steps (classification, extraction, drafting) done by a model with a person approving. |
| Speed to start | Days: sign up, configure, import. | Weeks to scope, then working software every couple of weeks. | As custom; AI features need evaluation data first, so they land in a later block. |
| Ongoing cost shape | Per-seat or per-module subscription, rising with headcount and add-ons. | Build cost per stage, then hosting and maintenance you control. | As custom, plus model usage billed on volume, kept in check with routing and caching. |
| Integration | Limited to the vendor's connectors and API. | Whatever your systems expose, built to your data model. | As custom. Models can also read unstructured inputs: email, PDFs, photos. |
| Ownership | Licensed. Data export on the vendor's terms. | Yours: code, infrastructure and data. | Yours, including prompts, evaluation sets and pipelines. |
| When it is the right call | A standard process, a small team and little to integrate. | The process is part of your edge, or the workarounds now cost more than a build. | A custom system where the volume of judgement work is the bottleneck. |
Custom platforms we have shipped
Three systems from our case studies page, with outcomes as reported by the clients.
DisabilityAssessments
An NDIS allied-health operations platform: onboarding flows, a central knowledgebase, automated matching and alerting, end-to-end invoice processing for providers and plan managers, automated reconciliations and secure patient-data storage. Client-reported outcome: 90% of operations automated.
Read the DisabilityAssessments case studyLuxpip
A FinTech trading business: design system, web and mobile apps, an admin panel and a custom B2B API, with agentic tooling for internal operations. Client-reported result: manual workload down 75% and a 40% scale-up.
Read the Luxpip case studyBilly
A custom platform with agentic workflows running through the core operations of the business, so routine work is carried by the system rather than added to someone's day.
Read the Billy case studyA custom software development company that also does the AI
One team from first call to production
A small senior team, remote-first across Australia and onsite where it helps, carries the system from scoping to deployment.
We build what we recommend
Strategy and engineering are the same firm, so a plan is never something another team must make work. If the honest call is to buy, the plan says so.
Priced stage by stage
Two free conversations, then Discovery from a fixed fee, Strategy priced upfront and Implementation per 12-week block. Exit at any boundary.
Built to be maintained by others
A mainstream stack, tests, infrastructure-as-code and documentation, so another team could pick the system up. AI goes in as a feature, not as a dependency on us.
Adjacent work we do
SaaS development
Multi-tenant subscription products built to sell to other businesses.
Custom SaaS platformsAI feature development
Adding AI to a product or internal tool you already run, without a rebuild.
Add AI to an existing productSoftware engineering consulting
Architecture review and senior engineering direction for systems you already run.
Architecture review and directionAI integration
Model outputs wired into Xero, HubSpot, Salesforce and your own databases.
Integrating AI with existing systemsCloud engineering
AWS account setup, migration, cost and the platform underneath the software.
Cloud consultingAI development
Applications where the model is the core of the system rather than a feature.
AI development servicesQuestions owners and operators ask
Six answered here. Bring the rest to a free call.
Book a free intro callIs custom software cheaper than a subscription?
Not at the start: a subscription costs less in month one and a custom build costs more upfront. Over time it depends on hosting, maintenance, any model usage and what it replaces in seats, add-ons and manual work. The Strategy stage puts numbers on that and makes the build, buy or defer call before anything is built.
How long does a custom software project take?
Discovery takes two to four weeks, and a Strategy stage, where one is needed, another two to four. Implementation runs in 12-week blocks with working software every couple of weeks; the number of blocks depends on how many workflows are in scope, and each ships something people can use.
Do we own the software?
Yes. The IP is yours from day one of any paid stage. Code lives in your repositories, infrastructure runs in your own AWS account or one we set up for you, and tests, infrastructure definitions and documentation come with the handover. No licence back to us, no mandatory retainer.
What if we want to change developers later?
You can, at any stage boundary. We build on a mainstream stack (TypeScript, React, Node or Python, Postgres, AWS), keep tests and infrastructure-as-code in the repository and document the architecture as we go, so a competent team can pick it up from the handover pack.
Where is the software hosted?
In your own AWS account, or one we set up for you, deployed in the region you need; for Australian data residency that is the AWS Sydney region. Where the account is yours, you hold the root credentials and billing relationship; we work through least-privilege roles you can revoke at any time.
How do you handle security and client data?
Authentication through Cognito or your existing identity provider, role-based access, encryption in transit and at rest, managed secrets, least-privilege IAM and audit logs on sensitive actions. For regulated data, such as patient records in our NDIS work, we design the data boundaries first and review them with you before the build starts.
Tell us which workflow hurts most
Bring the spreadsheet, the tool you have outgrown or the process nobody can describe. Two conversations cost nothing, and you see a written price for each stage before choosing the next.