Polymorphic Applied AI consulting

Polymorphic · Applied AISydney · Est. 2022

We find the most impactful AI solution for your business. And build a working system around it.

We're ML engineers with over fifteen years experience shipping production systems. We work with your leadership to understand your business, find where AI actually makes a difference, build that system, and run it live. What you get is a working part of the business, still running after handover.

LiveYou're watching a live variational autoencoder trained to reconstrucct damaged forms, finding a solution in real time, and rebuilding it piece by piece.

Experience15+ yrs

The seasoned experts who scope your problem are the same ones delivering you the final product.

EngagementEnd to end

Strategy, build, and production ownership in one team — first framing to live system.

First evidence6–8 weeks

A working system on your data, measured against the baseline you run today.

Measured onYour P&L

Margin, cycle time, capacity — measured success defined in your numbers, not just model metrics.

How we work

  1. 01

    Business case first, technology second.

    We make a point of getting a deep understanding of your business first. Once we find the highest-leverage problems that emerging technology can solve for you, only then do we build — whether that's a model, a forecast, an optimiser or a better process.

  2. 02

    The problem is not the problem.

    When projects stall, it’s usually for the same few reasons: the wrong problem, toy data, solutions not fit for purpose in the real world. These are all artefacts of poor planning. By getting this first step right, we cut out the bulk of project risk at the outset.

  3. 03

    A model is not enough.

    To move the needle, a solution has to be integrated into the ecosystem around it — the data, controls, evaluation and people in the loop. We deliver end-to-end solutions that fit into your systems, broken into clear milestones that limit the impact of unknown unknowns.

  4. 04

    Move fast, test early.

    We build the smallest credible system on your data, then make a clear go or no-go call. If it works, we fit it into your workflow and make sure someone owns it, whether that’s your team or ours.

Start here

Bring us your hard problems.

We will tell you what is feasible, what it is best suited to your needs, and the shortest path between the two.

Book a working session
Case studiesSee all
  1. 480×faster processingInsurance
  2. 2.3×viable strategies at launchGames
  3. ±12%annual load estimate vs meteredHousing

From our lab · live

NEM Radar

Our own forecasting model for Australia’s electricity market, running in the open: five-minute prices and the odds of a spike in all five regions, checked against the market operator’s forecast as it goes.

The record since August 2024, fed through a day at a time: each column is a day and each dot an hour, coloured where the price went above $300. Days with two hours or more above $300 are stamped with their peak, to two figures.