How it works
Free scope. Free build.
You only pay if it works.
Most AI consultancies sell six-figure feasibility studies. We don’t. We build the agent on one of your real workflows for free, hand it to you, and you choose how to run it. If it doesn’t work, no obligation.
The 6-step pipeline
Same shape for every engagement. Tier 1, 2, or 3 affects the build complexity — not the steps.
- 01
Scope-out call (30 min, free)
Microsoft Teams. We map one of your real workflows live — what inputs, what outputs, what 'good' looks like. By the end of the call we both know if it's a fit. No deck, no hard sell.
What you bring: 5 minutes of context on the workflow that's eating your week. What we bring: 8 years of production AI experience and a one-page recommendation.
- 02
Deeper scope-of-work (1-2 hours, free, optional)
If we agree it's worth building, we go deeper. Walkthrough of your tools, sample inputs, signed-off comparable outputs. NDA before this step if you need one.
This step is optional but recommended for Tier 2 / Tier 3 builds where data flow matters. Skip it if Tier 1 is the obvious shape.
- 03
Free agent build (1-3 weeks)
We build the Claude-native workflow on Anthropic's stack with AU data residency (AWS Sydney, Google Vertex Sydney, or Azure AU). You don't pay anything yet.
Most builds land in 1-2 weeks. We absorb the build cost because the goal is to prove fit on real data — not to sell you a feasibility study.
- 04
Delivery + decision call (60 min)
We hand it over and run it on a real claim/case/file with you. You see the math: hours saved, accuracy vs. signed comparable, cost per task. Then you decide.
If it doesn't work, no obligation. We share what we learned and you walk away with the audit. We've taken the risk, not you.
- 05
Three options to run it
PAYG (per-task — we host, you pay only when the agent runs), monthly hosted (we run it, fixed monthly fee, includes monitoring + updates), or buy outright (we hand over the codebase, you self-host).
Final pricing is set at the scope-out call — every engagement is custom-sized. The framework is public, the numbers are tailored to your volume.
- 06
We host and run it
If you go hosted: we operate the agent on Anthropic's frontier models, monitor cost + accuracy, ship improvements quarterly, and surface compliance audit trails on demand.
Long-term ecosystem: most clients stay hosted because the maintenance load is real. Anthropic ships new models every quarter and you want them on day one, not month six.
Why we can do free build when others can’t
Claude Code compresses build time. What used to take 6 weeks now takes days. We’re a Claude-native shop running the same tooling on our own internal 40-agent fleet — we eat our own dog food.
We're optimised for ecosystem capture, not project margin. Our model is to host long-term — that’s where the durable revenue is. The free build is the foot-in-the-door, not the loss-leader to a one-off project.
We pick the right A/A/A mode. Most AI projects fail because the buyer is sold Agency when they were ready only for Augmentation. We size the engagement to your readiness, not our revenue target.
Ready to scope one of your workflows?
30 minutes. Free. No deck, no hard sell. By the end of the call we both know if it’s a fit.