Monthly
Flat subscription per system
No capital project, no change-request billing. Cancel when it stops paying for itself.
Our forward-deployed engineers spend time on your floor, identify the process costing you the most, and ship a working system within weeks. Discovery costs nothing. You don't buy a project or carry a codebase — you subscribe to a private product we keep improving, with the numbers to prove it is working.
A large consultancy sells you a build, hands over a repository, and leaves you to maintain software you never wanted to own. We do the opposite: discovery is free, we build the product, host it, support it, and keep improving it — and you pay a monthly fee for as long as it earns its place.
No capital project, no change-request billing. Cancel when it stops paying for itself.
We visit, scope and tell you what is worth building before any money changes hands.
Support goes to the people who wrote it. No ticket tier, no account manager relaying it.
Every system reports the baseline it replaced and what it is saving now.
First visit to live system: about a month.
A few visits with the people doing the work. We watch the process, time it, and write down where it actually breaks — at our cost.
Starts within a weekOur team builds the system end to end — models, interface, integrations, hosting. Nothing lands on your engineering backlog.
Included in subscriptionLive with a small group within a month of the first visit, then tightened weekly against how your team really uses it.
Weekly releasesHosting, monitoring, support and continued improvement for a flat monthly fee. Your team uses it; we keep it healthy.
Cancel anytimeBefore anything ships we measure the process as it stands — cycle time, touches, rework, escalations. Every system then reports against that baseline continuously, and the numbers go to you, not just to us.
One monthly review, one page, the same metrics every time.
Forward-deployed engineers work from your floor to see how the job is really done, then replace the fragile manual paths with a product your team logs into.
Frontier lab APIs where capability decides it, self-hosted open-weight models where sovereignty or cost does. We choose per system and can run it in our Australian environment, your VPC, or your own racks.
The work that has to move something. Vision, control and scheduling systems that land on the production line and hold up across shifts.
Yield, scrap rate, changeover time and inspection on live lines.
Sovereign deployment inside accredited and air-gapped environments.
Asset inspection, outage forecasting and field reporting at scale.
Sensitive records handling under strict residency and consent rules.
Freight planning, exception handling and depot operations.
Remote monitoring, predictive maintenance and compliance reporting.
Names withheld under NDA
Vision inspection across two injection-moulding lines replaced a manual sampling regime that was catching defects a full shift too late. Live on one line in nine days; both lines within six weeks.
An open-weight model fine-tuned on a decade of case files, running entirely inside the agency's own environment, triaging inbound correspondence before a human sees it.
Dispatch exception handling moved from four planners and a shared spreadsheet to a system that escalates only what genuinely needs a decision — refined weekly since rollout.
[ Placeholder outcomes — replace with verified figures before publishing. ]
No — and that is the point. We retain the code and run it as a private product for you, which is why it costs a monthly fee instead of a capital project. You own your data, and you can export it in full at any time.
Two reasons. We use frontier AI tooling to supercharge our own engineers, so a small senior team ships production software in a fraction of the time a delivery pyramid takes. And because we host and maintain the product ourselves, you are not funding the codebase handover or the internal team you would then need to keep it alive.
The build itself, hosting and inference, monitoring and uptime, direct support from the engineers who wrote it, continued improvement, and the monthly efficiency report. Discovery is free and sits outside it entirely.
You give notice, we export your data and help your team return to the prior process or an alternative. There is no exit fee. If a system stops earning its keep, we would rather you said so.
Not unless you ask for it. Systems run in Australian regions, in your own environment, or on your hardware where the rules require it, with SOC 2 aligned controls and full audit trails.
Often it isn't. Around a third of what we find during discovery is better fixed by changing a process or an integration, and we say so in writing rather than selling you a subscription. Discovery is free, so that outcome costs you nothing.
Discovery visits usually begin within a week of a first conversation, and a first product is typically live within a month of that.
A paragraph is enough to start. An engineer — not a salesperson — replies within two business days.
hello@micg.ai