AI consulting · Australia and New Zealand
AI consulting and automation that pays for itself
We find the workflow that costs your team the most hours, build an AI system that does it reliably, and hand it over running in your own cloud account. Fixed price, senior engineers, no lock-in.
Where AI earns its keep
Not the most impressive demo. The work that eats the most hours, happens every day, and follows a pattern a person could explain.
Documents people retype
Invoices, applications, claims, contracts, purchase orders and submissions. We extract the fields into your systems, with a person reviewing anything the model is unsure about.
Questions that interrupt your experts
Internal assistants over your policies, product documentation and past tickets. Every answer cites its source, so staff can check it in one click.
Inboxes and queues
Triage, routing and drafted replies waiting for approval. Your team handles the exceptions instead of the volume.
Reporting and admin
Weekly summaries, CRM updates, meeting notes turned into actions, and the reconciliation nobody wants to do on a Friday.
What you get
A system your team can run, measure and change without us.
- A working AI system deployed in your AWS, Azure or Google Cloud account, not ours
- An evaluation suite built from your real examples, so accuracy is a number you can track
- Human review built in where it matters, with confidence thresholds you control
- Guardrails for personal information, access control and a full audit log
- A cost and usage dashboard with budget alerts
- Documentation, a runbook and recorded walkthroughs
- A handover training session for the people who will own it
How it runs
Each stage has a fixed price and a definite end. If the pilot doesn't hit the accuracy target we agreed, you've spent three weeks finding out, not six months.
Discover
We map the candidate workflows, measure the hours they take and rank them by return and effort. You get a written report and a fixed quote for the pilot.
One week · $4,900Pilot
We build the system on a sample of your real data, measure it against the examples we collected, and show you the results every week.
Three weeks · from $18,000Roll out
Integration with your systems, monitoring, access control and staff training. Then we hand it over.
Six to ten weeks · fixed quoteHow we make AI work in production
Most AI projects stall between a good demo and a system people trust. These habits are how we close that gap.
Evaluation before prompts
Before we write a prompt we collect 50 to 200 real examples and agree what a correct answer looks like. Every change is scored against them.
People stay in the loop
Uncertain cases go to a person by design. The system gets faster as confidence grows, without ever being trusted blindly.
The right model for the job
We choose models on accuracy and cost for your task, and keep the system able to switch. Where data residency matters we use Amazon Bedrock in the Sydney region.
Unremarkable infrastructure
Serverless, logged, monitored and alerting on spend. The clever part is the workflow, not the plumbing.
Tools and platforms we use most
Pricing
Fixed prices, agreed in writing before any work starts.
| Option | Price |
|---|---|
| Discovery sprintOne week. Workflow mapping, data review, ranked opportunity report and a fixed quote for the pilot. | $4,900 |
| AI pilotThree weeks. A working system on your real data, with an evaluation report. | from $18,000 |
| Production roll-outSix to ten weeks. Integration, monitoring, security review, training and handover. | fixed quote |
| Ongoing supportA named engineer for improvements, monitoring and new use cases. | from $3,500/mo |
Most pilots land between $18,000 and $45,000. Model and hosting costs are billed directly to your cloud account, and we estimate them per transaction during the pilot. Registered startups under two years old get 15% off their first engagement.
What this looks like in practice
An illustrative example of a typical engagement.
Common questions
Will AI be accurate enough for our work?
We find out with your data, not a vendor's demo. During the pilot we measure accuracy against real examples and agree a target up front. Tasks that need a person's judgement keep a person in the loop.
Is our data used to train AI models?
Not by default. Under their commercial terms, Amazon Bedrock and the major model APIs do not use your inputs to train their models. We confirm the settings for your account and write down where your data goes.
Do we need clean data before we start?
Usually not. Most document and knowledge use cases work with what you already have. The discovery sprint tells you plainly if a data clean-up needs to come first.
What does an AI system cost to run?
We give you a cost per transaction during the pilot and set budget alerts in your account. For most small and mid-sized businesses, model costs are small next to the staff time the system saves.
What does the Privacy Act mean for AI?
If you use personal information in automated decisions that significantly affect people, the 2024 amendments require you to say so in your privacy policy from 10 December 2026. We design systems so you can explain what data is used and how decisions are made.
Often paired with
Data and analytics
A governed data platform that brings Xero, Shopify, HubSpot and your databases together, with dashboards your team trusts.
Learn more →Essential Eight · Privacy Act · SOC 2 prepSecurity and compliance
Practical security for growing companies: Essential Eight alignment, Privacy Act readiness and the evidence enterprise customers ask for.
Learn more →AWS · Terraform · FinOpsCloud and DevOps
AWS architecture, migration and cost reduction, with infrastructure as code, CI/CD and monitoring set up once and explained properly.
Learn more →Find the workflow worth automating first
A one-week discovery sprint for $4,900 gives you a ranked list of opportunities, a realistic cost for each, and a fixed quote for the pilot.