How we build
We build around your workflow, then run it.
Seven steps, from mapping your workflow to choosing the model, measuring it on your real work and running it every month from go-live.
01
Start with the workflow
Every system we build starts with the work: the workflow that eats hours, creates avoidable cost or carries risk your team should not have to carry. In Discovery we map how the work moves today, who touches it, where judgement matters and what changing it is worth. Only then do we talk about technology.
Once we understand the workflow, we know what the model has to be good at and what it is allowed to get wrong.
02
Choose the right model
There is no single best model. Depending on the job, the right answer might be a frontier model, an open-weight model hosted privately, a smaller model specialised for one task, or a model your organisation has already approved. We are independent of any one provider, and we weigh every candidate against the same criteria.
What we weigh
- Quality on the task
- Cost per unit of work
- Latency
- Privacy
- Data residency
- Reliability
- Tool use
- Context requirements
- Change frequency
- Review requirements
03
Adapt the model where it earns its keep
We start with the most cost-effective and controllable rung that works, and move up the ladder below only when the evidence shows it has reached its limit.
Fine-tuning is on the table when a workflow, its data and its economics justify it. It is an engineering option we hold, and it is rarely the first one we reach for.
- 01Prompt and context design · The most efficient first step. Clear instructions, the right context and the right examples.
- 02Retrieval and structured context · The system looks things up in your data instead of guessing from memory.
- 03Workflow controls · Validation, checkpoints and review boundaries around the model, so the workflow stays correct even when an output is wrong.
- 04Fine-tuning and adaptation · Training an existing model on your work, where the volume and the value justify it.
- 05Full custom training · The top of the ladder, for the rare case that needs it.
04
Measure against your work
We measure each model on examples from your own workflow: the everyday cases, the edge cases that hurt, the policy constraints it must respect and the numbers your business already watches.
A route that falls short on those examples does not go live, whatever it costs. After go-live we keep measuring quality against the same examples.
05
Build securely for production
A system in production has to be secure, governed and meet your compliance obligations. We build the controls below into each system from the start, with auditability and clear accountability.
- 01Hosting on Microsoft Azure in Australia
- 02Data handling and residency matched to your obligations
- 03Monitoring and alerting on quality, cost and behaviour
- 04Versioned models, prompts and datasets
- 05Access control and permissions
- 06Failure handling and rollback
- 07Human review points where the work needs judgement
06
Route requests deliberately
Routegate is the part of our engineering that picks which approved AI model handles each request, inside the systems we build and run.
The rules come before cost. Routegate only uses models we have measured and approved, and it will not pick a cheaper one that falls short on quality or breaks a rule. It records why it chose each model, keeps your data apart from its monitoring logs, and stops when a required setting is missing or out of date.
You cannot buy Routegate on its own, and there is nothing for you to operate.
07
Operate the system
Go-live is where Run and Maintain begins. We monitor every run, review exceptions with your team, track quality against the original examples, watch for drift as models and providers change, and reassess the costs as usage grows. When a better or more cost-effective approved model becomes available, we review the change with you and move to it.
We stay accountable for the result for as long as we run it.
Questions
Asked often, answered plainly
Which AI models do you work with?
We choose the model to fit the work: frontier models, open-weight models or smaller specialist models, including models your organisation has already approved. We are independent of any single provider.
Can you fine-tune a model for our business?
Yes, where the workflow, the data and the economics justify it. We first check whether better context, retrieval or workflow controls solve the problem more simply.
Do you sell an AI platform?
No. We build each AI system for your work, then run it for you. You get a working system inside your business and a team accountable for keeping it working, with nothing for you to configure.
What is Routegate?
Routegate is the part of our engineering that picks which approved model handles each request, inside the systems we build and run.
What happens after go-live?
Run and Maintain begins, for a monthly fee. We monitor every run, review exceptions with your team, manage model and provider changes, and make the small changes you ask for each month.
Where to start
See it applied to your business.
Every build starts with Discovery, a paid audit of how your workflow runs today. At the end of week one we demonstrate what we have found. If a build would not pay off, we stop work and you pay nothing.