Guides · AI
AI readiness, honestly assessed.
Most AI-readiness content is a sales funnel with a quiz attached. This one is the actual questions we ask in week one of an AI Transformation - including the ones that end engagements early, which is the point.
Readiness is four questions
- Is the workflow frequent, boring and measurable? AI pays for itself on volume. A task done twenty times a day by five people is a candidate; a quarterly judgment call is not.
- Does the data exist, and can you legally use it? The documents, records or messages the AI must read - do they exist digitally, with enough history, and with the rights to process them? “It is in people’s heads” means the project is knowledge capture first, AI second.
- Can a mistake be caught? Every model is sometimes wrong. Readiness means a checkpoint where a human (or a verifiable rule) catches the wrong answer before it costs something - if no such checkpoint can exist, the workflow is not ready, whatever the demo shows.
- Will anyone own it? An AI workflow needs a person who watches its metrics, feeds its evaluation suite, and decides when it drifts. Unowned AI degrades silently.
Signals you are ready
- You can name the workflow, its weekly hours, and its current error cost without a study.
- A baseline can be measured in days: time per task, volume, quality.
- Ten to fifty real examples can be assembled for an evaluation suite this week.
- Someone already complains about the task by name - adoption lives where pain is felt.
Signals you are not (yet)
- The goal is “an AI strategy” rather than a named workflow.
- The data lives on paper, in silos with no export, or under rights nobody has checked.
- The plan has no human checkpoint because “the model will be accurate”.
- Success cannot be stated as a number anyone currently measures.
Not-yet is a fine answer: the pre-work (digitisation, data rights, baselines) is ordinary software work, and doing it first is what makes the AI part boring - our favourite property.
Want the week-one audit for real?
Bring one named workflow. We will run these questions on it - free, and honest about not-yet.
Related: AI Transformation · AI Product Development · All guides