AI that handles sensitive data
De-identification before the model, fail-safe leak checks, and an architecture your compliance team can actually approve.
AI SaaS for regulated industries
We build AI products for industries that can’t afford to get it wrong.
Software that used to need a team of engineers — designed, built and run by a small senior studio. Two of our platforms are in production today.
- 2 platforms in production
- Multi-tenant
- AWS-native
- 25 years in regulated technology
The difference
Most AI companies have a demo. We have two platforms in production.
The distance between a working prototype and software a regulated business will actually run is where most AI projects die. Tenant isolation. Audit trails. What happens when the model is wrong. Where the data lives and who can reach it. Billing, roles, admin tooling, support. We have built all of it, twice, and we ship to production — not to a slide deck.
What we’re good at
Four things we do well
Multi-tenant from day one
Isolation, roles, per-tenant branding and central administration. Retrofitting this later is a rewrite.
Documents and unstructured data
Extraction, structuring, and turning what a business already has into something it can use.
AWS architecture and delivery
The infrastructure under both platforms — containers, managed Postgres, CDK, CI to production.
Selected work
Two production AI platforms
Built for Planfocus Consulting, an Australian advice-technology firm. Both are multi-tenant, both are live, and both solve problems that recur in every regulated sector.
Case study 01
AdviceStudio
Turns a dense 40-page compliance document into a client-ready interactive presentation in minutes — and the client’s personal data never reaches the AI.
A dedicated redaction engine strips every identifier before a single word is sent, then re-scans the result. Anything that slipped through is a hard stop.
Python
FastAPI
PostgreSQL
ECS Fargate
Case study 02
Coremetryx
Enterprise business intelligence, wrapped so a small business can use it — multi-tenant, branded dashboards built on Amazon QuickSight.
Complete data isolation per organisation, six granular roles, and dashboards embedded securely into a custom front end.
TypeScript
React
AWS CDK
QuickSight
How we work
Small team, production software
01
Senior only
No layers, no handoffs to juniors. The person who scopes the work is the person who builds it.
02
Production, not prototypes
We measure success by software running in front of real users, with billing, monitoring and support.
03
AI-assisted engineering
It is how we deliver a team’s output. It is also why our estimates look different from everyone else’s.
04
You own it
Documented handover of the code, the infrastructure and the decisions behind both.
Tell us what you’re trying to build.
Or what isn’t working. We reply within one business day.