Show us the mess
Don't tidy up first. The actual state of things is the diagnostic, and a cleaned version hides exactly where the process breaks.
Method
Assess. Optimise. Then automate.
In that order, without exception. Automating a bloated process doesn't fix it, it buries it somewhere harder to see. So we don't quote a build before we've looked at the work.
Why the order matters:if you automate a twelve-step approval that should be six steps, you haven't improved anything. You've made the extra six steps permanent and harder to see, and you've paid for the privilege. The optimisation pass costs a few days and it is the difference between a system that holds up in year two and one that quietly gets worked around.
The engagement
Four stages, and what you actually get from each.
- 1
Assessment — 7–10 business days, $2,500
We follow one finance workflow end to end, then score its opportunities by value, volume, readiness, effort, control risk and provability. You get the map, baseline, control design, economics and acceptance tests. It is yours whether or not you hire us, and the fee is credited to a build started within 30 days.
Founding-pilot terms: the first two qualified partners can buy the assessment for $1,500 with deeper baseline access and permission to request a case study after value is demonstrated.
- 2
Optimise — before anything gets built
Most firm processes carry steps that exist because somebody left in 2021. We cut those first, on paper, with your team. A twelve-step approval that should be six gets rewritten before a line of automation is built.
This stage is unglamorous and it is the reason our builds hold up. It is also where we occasionally discover that the answer is a process change and no software at all, which we will tell you even though it costs us the project.
- 3
Build — fixed scope, two to six weeks
We build the clean version, connected to the systems you already use. Every automation is versioned and tested like software, documented as we go, and deployed on infrastructure you own. Judgment stays with a human by design: agents prepare the work and flag what's odd, people decide.
We sequence so the highest-value system ships first and starts paying for the rest, rather than a six-month programme that delivers nothing until the end.
- 4
Support — monthly, and optional
Systems break when a bank changes an export format or a portal changes a field. We monitor them, fix what breaks, and ship one improvement a month. You get a report with the numbers that matter. This tier is optional: you own the system, documentation and test assets, and your team can operate them without us.
How we decide
Three tests before we build anything.
If a proposed automation doesn't clear at least one of these against a real number, it doesn't go in the roadmap. This is why our proposals are shorter than you expect.
Does it make money?
Work won that would otherwise have gone elsewhere, cash recovered, capacity to take another client. Speed to lead and receivables usually clear this one alone.
Does it save meaningful time?
Not fifteen minutes here and there. Hours per month, from named people, on work that is genuinely repeatable. If we can't name the hours, it isn't real.
Does it improve quality?
Fewer errors reaching a client, a process that doesn't depend on one person's memory, an audit trail that exists. Sometimes this alone justifies the work.
The fastest way to waste money on automation is to build something impressive that nobody needed. We'd rather deliver three things that matter than eleven that demo well.
Working with us
What we ask of you, and what we won't do.
Give us one decision-maker
Someone who can say yes to a process change. Automation projects stall on organisational agreement far more often than on technology.
We won't automate judgment
Estimates, unusual transactions, anything requiring a decision about treatment. Those stay with your team, and we design the queues that surface them.
We won't sell you a headcount cut
In practice people move to work that needs them rather than leaving. We'd rather say that upfront than have you discover it after the invoice.
Next step
Start with the assessment.
One finance workflow, 7–10 business days, $2,500 and credited to a build started within 30 days. You keep the blueprint either way.