Enterprise & local AI
If your AI setup is costly, slow or difficult to manage, start with the job it needs to do. We can assess an existing model or help your team run an agreed task on its own infrastructure.
We agree the task, infrastructure, responsibilities and support before delivery. Larger integrations and technology licensing are scoped separately.
Ways we can help
Test a job on an agreed compatible computer or approved reference machine. Get a local demonstration and a report on what worked and where it failed, so you can decide whether to deploy, make changes or stop.
Illustrative example: Can a private procedure assistant run on the workstation you already own?
Up to 10 business days
A$5,280 including GST (A$4,800 ex GST).
Set up one agreed task, test it with examples from everyday work and give your team the instructions to run it and recover when something fails.
2–4 weeks
Scope and fee agreed in writing before work starts.
If your current setup is costly or slow, tell us the model, hardware and job it needs to do. We’ll check whether conversion is supported and worth comparing with what you use now.
See a model conversion resultAgree one model, conversion or architecture question, how we’ll test it and how the work is funded. Keep the results even when the idea doesn’t work.
6–12 weeks
Rights, scope and funding agreed before work starts.
Before we start
We’ll agree the task and hardware, then compare quality, speed and resource use. You’ll get the results, including failed tests and anything we couldn’t measure, before deciding what to change.
Confirm the models, data and licences, who can approve access and what may be published.
Agree real examples, quality and resource limits, a reference to compare against and what should happen when it fails.
Agree who operates and supports it, how to recover, who approves changes and how you can leave the arrangement.
Talk to us
Tell us what needs to work, what you use now and who will use the result. We’ll come back with an approach and what we’d need to check before starting.