A practical checklist to determine whether a process is ready for automation, AI or preliminary standardization.
What readiness means
A process is ready when its objective, inputs, outputs, responsibilities and exceptions are stable enough to become explicit controls. A manual activity is not automatically worth automating. Volume, risk and data quality determine the real opportunity.
Readiness is not binary. Data collection may be ready for automation while the final decision still requires people, or AI may assist operators without taking autonomous action.
Process and data checklist
Confirm that an owner exists, the beginning and end are recognizable, and the main variants are known. Then verify source accessibility, completeness and freshness. Data must have shared meaning: fields with the same name across systems may describe different concepts.
Measure how much work takes place outside formal systems. Local spreadsheets, messages and email often show that the official model does not cover the real process.
Risk and governance checklist
Define who approves the automation, which actions require review and how each decision is recorded. AI outputs need evidence, confidence rules and fallbacks. Personal data, intellectual property and contractual constraints belong in the initial assessment.
Ask what happens when a source is unavailable or an output is ambiguous. Exception handling is part of the product, not a later enhancement.
Scoring and the next decision
Score clarity, data, stability, volume, risk and sponsorship. A low score identifies preparation work; a high score authorizes a pilot, not an immediate rollout.
Close with a pilot case, baseline metric and stopping criterion. This keeps the experiment controlled and prevents a technical demonstration from being mistaken for full adoption.
Next step
Before choosing a tool, assess the process with the Process Readiness assessment. For complex initiatives, explore the AI Process Intelligence method and ControlRoom use cases.
Frequently asked questions
Can an unstable process be automated?
Only its stable parts. Frequent exceptions must remain visible and explicitly managed.
Which metric should come first?
Use the metric connected to the problem: cycle time, waiting, rework, error or decision latency.