Tool
Workflow check: is this process right for your first AI agent?
Ten questions about one specific process. You see how it scores on value, feasibility and ownership, and what has to be settled before a build.
Answer the ten questions for one specific process, not for the operation as a whole.
Result
0 of 10 answered
The result appears once all ten questions are answered.
Three lenses, the same as in the assessment
Value asks whether the process has enough cases, enough effort per case and enough consequence when it goes wrong to carry a build. Feasibility asks whether cases arrive digitally, whether the context sits in systems, whether the rules are written down and whether there are interfaces. Ownership and risk asks whether someone in the business wants the outcome, whether human release is accepted and whether the process decides about people.
A process has to pass all three. High value does not make up for missing interfaces, and good data does not replace an owner.
Two answers that block everything
If nobody in the business is measured on the outcome, the check recommends no build, however good the rest looks. Without that person you get a demo that nobody runs after the pilot.
If the process decides directly about people, for example applications, performance or creditworthiness, it can count as a high-risk system under the EU AI Act. That does not rule out a build, but it shifts effort and timeline. For a first workflow it is rarely the right choice.
What comes after the check
A strong candidate needs a baseline next: real case counts, measured handling times and an error rate from the last twelve months. With those numbers the business case can be worked through properly in the ROI calculator.
A candidate after groundwork first needs work on its weakest lens. Usually that means writing down the ten most frequent decisions with their rule, or checking with IT which systems have an interface.
Questions