Custom enterprise software Software you own, not software you rent.

Eleven things happen on a platform build. Here they are, in order.

This is the part most agency pages keep vague, and it is the part that decides whether you end up owning an asset or inheriting a dependency. So it is written down.

1. 01

   Map what you actually run first

   Every system, every integration, every spreadsheet doing load-bearing work, and what each one costs per year. On one shire engagement that audit found nine disconnected vendors before a line of code was designed.

2. 02

   Draw the data model before the screens

   One auditable model the whole platform reads from, rather than a module per department stitched together later. This is the decision that cannot be reversed cheaply, so it gets made first and in the open.

3. 03

   Choose boring technology on purpose

   TypeScript and React on the front, .NET 8 or Node.js on the back, PostgreSQL underneath. No bespoke languages, no framework nobody else hires for. The point is that somebody who has never met us can maintain it.

4. 04

   Build the hardest module first

   Not the easiest one. On larger programmes it becomes a free working prototype you can use before committing to full scope — if the hardest part works, the rest is scheduling.

5. 05

   Integrate two-way, not one-way

   Oracle Fusion, TechnologyOne, Civica Authority, Aurion, Workday, Microsoft 365, SAML and OIDC. Data flows back into the systems you are keeping, so the new platform is additive rather than a second source of truth.

6. 06

   Put AI where it removes typing, not judgement

   Variance commentary, service-request triage, accounts-payable coding, anomaly detection and natural-language questions over your own data. It runs in the Sydney region, and client data does not train external models.

7. 07

   Have a second person check everything

   Code review, migration rehearsal and release checks all carry two names. On a finance platform a single unreviewed migration can reach a budget cycle, which is exactly why nothing ships with one.

8. 08

   Host sole-tenant, in Australia

   AWS Sydney primary with Melbourne disaster recovery, in your own tenancy. Customer-controlled encryption keys, and the audit log forwarded to your security systems rather than kept in ours.

9. 09

   Lodge escrow from sprint one

   Source code held with ECI Australia. You can bring operations in-house or move to another provider without a re-implementation, and that is true from the first sprint rather than from go-live.

10. 10

    Write the runbook for a stranger

    Deployment, backup, restore, key rotation, on-call escalation and the known rough edges, written for somebody who has never spoken to us. A handover that depends on a conversation is not a handover.

11. 11

    Operate it, if you want us to

    Australian-staffed support against the uptime commitment, with a fixed reporting rhythm. If you would rather run it yourselves, the escrow, the keys and the runbook already make that possible.

From the ecosystem audit to the runbook.

1. Week 0

   The ecosystem audit

   We map every system, integration and spreadsheet in scope, with its annual cost and its owner. You get that map whether or not you go ahead with us — it is useful on its own.

2. Weeks 1–3

   The architecture

   Data model, integration surface, hosting topology, identity and access, backup and recovery targets. Written down and reviewed with your technical people before anything is built.

3. Phase 0

   A working prototype

   On larger programmes we build the hardest module as a free working prototype. You use it with your own data before either side commits to the full scope.

4. Sprints

   Build in modules

   Each module has its own acceptance window and its own review. Nothing merges with one name on it, and nothing reaches a live environment without a rehearsed migration.

5. Go-live

   Escrow, keys, runbook

   Source-code escrow lodged, encryption keys handed to your administrators, audit log forwarded to your security team, and the runbook walked through with the people who will hold it.

6. From there

   Operations or handover

   We run it against the uptime commitment with a fixed reporting rhythm, or we hand it over. Both were designed for from the first sprint, so neither is a renegotiation.

Nothing merges with one name on it, and nothing is handed over without the source, the keys and a runbook written for somebody who has never met us.

A platform build usually starts in one division and pulls in a second.

Most engagements start in one division and pull in a second. All five divisions sit side by side on one page, if you would rather start there.

• 01MarketingPerformance media, brand, growth.
• 03Public SectorCouncils, Commonwealth, cultural institutions.
• 04ResearchTrauma-informed surveys, sovereign data.
• 05CreativePlace branding, public art, civic publishing.