The problem
Everyone knows the process is slow, but nobody can say which step costs the most. Without a baseline, any automation looks like an improvement.
The exceptions are invisible until the automation meets them. A process that looks like ten rules is usually ten rules and forty exceptions.
The tool decision is being made before the process decision. Buying a workflow platform will not fix a process that has three unnecessary approvals.
The team has no shared picture of the workflow, so every conversation starts from a different version of it.
What you get
Outcomes
- A measured baseline: hours per week, volume, cycle time and error rate.
- An exception catalogue with the volume and handling cost of each case.
- A costed automation design, including the integration work and the human review step.
- A clear decision: automate, simplify first, or leave it alone.
- A short prototype or a tested prompt/extraction step where the data allows it.
Deliverables
- A process map from trigger to outcome, with owners and systems.
- A time study based on real samples, not estimates from a workshop.
- An exception catalogue with frequency and handling cost.
- An automation design: trigger, AI step, validation, write-back, human review and failure path.
- The integration list, including what has an API and what does not.
- A cost model with hours saved, implementation cost and payback period.
- A decision record and a proposal for the build, with a kill clause.
Fit and boundaries
This is a good fit when
- One workflow with real volume and a clear owner.
- Repetitive steps that follow rules most of the time.
- Access to real samples and the people who do the work.
- A team willing to hear that the process should be simplified first.
This is not a fit when
- A request to "automate everything" with no single process in scope.
- A process that changes every week or has no stable owner.
- No access to real samples. The audit would become a workshop.
- A decision that legally requires a human judgement with no AI involvement.
How it works
- 01
Pick one workflow
Not a department. One trigger, one outcome and one owner. If the scope needs a diagram with three swimlanes and two exceptions, it is too big for a five-day audit.
- 02
Watch the real work
I look at real samples and time the steps with the people who do them. The gap between the documented process and the actual process is where the hours hide.
- 03
Quantify the exceptions
Each exception gets a frequency and a handling cost. That is what decides whether the automation is a straight-through flow or a human-in-the-loop tool.
- 04
Design, price, decide
You get an automation design, an integration list, a payback estimate and a recommendation. Sometimes the recommendation is to remove a step instead of automating it.
Pricing and timing
Fixed-price engagement
€1,500 to €2,500 5 working days · EUR, VAT not applicable at present
Fixed price, five working days. Credited against the build if you continue within 30 days.
- Written scope and acceptance criteria before work starts.
- 40 to 50 percent deposit reserves the calendar slot.
- Milestone invoices and 14-day payment terms.
- A checkpoint where either side can stop with paid work settled.
Questions
What counts as one workflow?
One trigger and one outcome, such as "an invoice arrives and is approved or rejected" or "a support email arrives and is routed to the right queue." If it needs two departments and three handoffs, split it.
How do you measure the hours?
I time real samples with the people who do the work, then multiply by volume. The output is a range with the assumptions written down, not a single optimistic number.
Do we need to buy a tool first?
No. The audit decides whether you need a tool, an integration or a simpler process. Starting with a platform tends to produce an expensive workflow that mirrors the old one.
What if the answer is not to automate?
Then the audit says so. A removed approval or a better form can be cheaper than an AI step. The report is still useful because it shows where the time actually goes.
How long does the audit take?
Five working days for one workflow, assuming access to the team and the samples. The proposal for the build follows within two business days.
Page last updated: October 2026