Acknowledgement of Country
We acknowledge the Traditional Custodians of Country throughout Australia and their connections to land, sea and community. We pay our respect to their Elders past and present and extend that respect to all Aboriginal and Torres Strait Islander peoples today.
Artist Partners is an Australian data and AI engineering firm. We design and build the systems that make data useful — platforms, models, semantic layers and the software on top of them — working inside our clients’ teams.
Where our team has worked
What we do
Four capabilities, drawn from what we have shipped.
01
Reproducible from source, not assembled by hand.
Databricks and Unity Catalog. Medallion architectures. Ingestion from the systems that run a business — SAP, plant MES, CRM, ad platforms. Terraform, OIDC and CI/CD, so the platform is reproducible.
02
Measured against the method it replaces.
We establish the baseline first, then build against it. On a recent manufacturing engagement we replicated the existing rule-based estimate across historical orders before training anything, so the comparison held.
03
So the business can ask its own questions.
Connecting activity to outcome across dozens of systems, then modelling it so someone who does not write SQL can ask a question in English and get a reliable answer.
04
Software we design, build and run.
Multi-tenant platforms, conversational interfaces, agentic workflows. We built ANAMA and we operate it.
How we work
Our people join your standups, your repositories and, when it matters, your factory floor. On a recent engagement we spent time on the floor at three plants in three states before designing a single feature.
Our deliverables are repositories, pipelines, models and running software. We write a document when a document is the clearest way to explain something.
The people who scope your engagement are the people who write the code. Anson MacDonald, our CTO, is completing a PhD in statistics at UNSW. Sarthak Das, our Head of AI, holds two AI-related patents.
We work inside your platform and your repositories, using patterns your team can extend. On a recent engagement the SAP extraction pattern was designed for reuse across the client’s wider Databricks programme.
Every phase is gated, and the first one is free. Before a build starts we agree the evaluation dataset, the baseline calculation and the success threshold in writing.
Notes
Written by the people doing the work. All notes.
Measurement · 8 August 2026
Connecting spend to revenue means an unbroken chain of joins. Where it breaks you get a number nobody can act on.
Method · 7 August 2026
A holdout score says a model agrees with its own test data. It says nothing about whether it beats the spreadsheet it was built to replace.
AI systems · 7 August 2026
Getting a model to write a decent client update is the easy part. The hard part is the two minutes a human spends approving it, and the clock governs those.
Architecture · 7 August 2026
The numbers everyone trusts often sit behind a desktop add-in or a licence nobody bought. That should change the architecture. It should not stop the build.
Partners
