Sprinkling Act
You can’t govern what you can’t see.
See where AI costs rise, and where work quietly shifts.
Two third-party measures · neither on an SME sample
Your AI is already working. Can you see the work around it?
DOES THIS SOUND FAMILIAR
The deployment worked. Then the work around it started.
The extra human work may well be justified by what it produces. The problem is deciding without seeing it.
WHY NOW, AND NOT IN SIX MONTHS
Four moments give the question a date.
AI spend has risen over two consecutive quarters and the outcome metrics have not moved. The spend has a trajectory. The results do not.
A board, investor or budget review is coming, and there is no defensible answer to “why has this stalled?” that does not come from the people who built it.
Something was built in-house and the moment to reinvest or cut it has arrived. The survey costs a fraction of the decision it informs.
A European customer has sent an AI governance questionnaire, or a contract renewal carries one.
If none of these four is ahead of you, the survey can wait, and we will say so rather than sell you one.
WHO THIS IS FOR
This is for you if
- Your AI spend is going up and your AI outcomes are not.
- You built something internally and cannot tell whether to double down or cut it.
- A board, investor or budget review is coming, and no defensible answer with it.
- A European customer has sent you an AI governance questionnaire.
This is not for you if
- You are still deciding whether to use AI. We reconstruct what exists; we do not draw roadmaps for what does not.
- You want a maturity score, a benchmark, or a deck that hands you back your own answers.
- Nothing depends on the answer in the coming months. The findings would be dated before anyone used them.
Check whether this is your situation Measure your exposure →
AN INDEPENDENT VIEW
An independent view before the next decision.
WHAT EACH SIDE KNOWS
Your provider knows their tool better than anyone. Your teams know their own work. Both are necessary sources, and we start from them.
WHAT IS MISSING
Nobody's role is to bring the two together across the whole workflow. That is the work we do, so that everyone is looking at the same picture when the decision comes up.
WHAT IT CHANGES
A single reading that your provider, your teams and your management can go through together, and use as a common starting point for what comes next.
Nothing is sold after the report: no tool, no license, no automation, and no reselling of anyone's platform. That is what makes the finding checkable, not a promise of neutrality.
Compare it with your current setup Request a scoping call →
ONE VIEW
What the analysis brings into one view
Some of this you already track. Some of it sits in several places at once. The point is to look at all of it together before deciding what happens next.
EVIDENCE STATUS
Every finding rests on the procedure as written, on the material we were able to examine, or on an inference we make explicit. Each one carries one of these four statuses, and a gap is only presented as established once a second, different-type source confirms it.
- ScopingThe workflow is named and the scope agreed in writing
- InterviewsWith you alone — your teams come in at Full Map
- ElementsWhat you provide, read and cross-checked against what was declared
- ReadoutWritten report, then sixty minutes together
Bounded scope · artifacts over memory · limits stated
WHAT THE USE PRODUCES
OUTPUT
What the tool produces
volume · turnaround · Read from activity records
Volume handled and time saved on the task itself. This is usually the number the deployment was justified on.
DIRECT COST
What the tool costs
licences · API · seats · Read from invoices
Subscriptions, consumption and seats. Visible on an invoice.
SETUP
What it took to start
integration · training · Read from project records
Integration, configuration and training. Counted at launch, and worth revisiting when the workflow has moved since.
WHAT IT DRAWS ON
SUPERVISION
Checking, correcting, re-doing
Someone reads the output before it goes out. How often that happens, how long it takes, and what happens when they send it back is worth putting a number on.
HAND-OFFS
What is lost between two steps
Context and ownership travel between people and systems. Most of the friction sits on those transitions.
DEPENDENCY
What rests on one person
Workflows often rely on one person who catches what the tool misses. Knowing how much rests on them is useful well before they are away.
RECORDED FOR EACH HUMAN TOUCHPOINT
Six dimensions. One bounded scope.
Moment
Before, during, after, or on exception only
Authority
Decide, approve, be able to halt, or only observe
Frequency
Every action, a sample, a threshold, or never
Information
Full context, summary, alert, or not enough to act
Responsibility
Explicit, shared, implicit, or unassigned
Recovery
Immediate fix, rollback, escalation, or none
ISO 42001 and the AI Act describe what oversight should look like. Neither tells you where the humans in your workflows actually stand today.
CLARITY
What this is not.
Not an audit, not a monitoring tool, not a reseller.
An agreed scope, dated findings, and the open questions named.
Then, downstream
Most work sold as governance assumes you already know how the work runs today. That is where our AI Act corpus belongs: after the survey, not before it.
Dated, not permanent
A survey holds for a scope and a date. Workflows move, people leave, tools get added. We say from what point a finding stops being reliable.
One decision. One bounded survey.
Tell us which decision is coming and what already runs. If there is no fit, we will say so in that first conversation.
Request a scoping callRegulatory signals, when they happen.
Unsubscribe anytime · no third-party sharing. Need the same workflow re-examined? Periodic re-analysis.