AI
Invest when you have a specific, repeated task involving unstructured information, and the data that task needs already exists and is accessible. Investing because AI is expected of you produces pilots that never reach production.
'We should use AI' has no success criterion. 'We spend twelve hours a week reading enquiry emails and routing them' does: it names the work, the volume and the thing that would change. Projects framed the second way survive contact with reality far more often.
AI applied to your business needs access to your business's information: documents, records, history. If that material is scattered, contradictory or locked in systems nobody can query, the first project is data access, not AI. Most stalled initiatives stalled here.
Drafting a reply a person approves, classifying an enquiry, summarising a call: errors are visible and cheap. Irreversible or regulated decisions are the wrong place to begin, whatever the demo suggested.
It is legitimate to conclude not yet.
If the process is unstable, the data is not ready, or nobody owns the outcome, waiting is a decision rather than a failure, and considerably cheaper than a pilot that quietly ends.
You need the data the specific task requires to be accessible and reasonably consistent. That is a much smaller job than a full data programme, and it is where most stalled AI projects actually stopped.
A narrow, repeated task with unstructured input and a human check: classifying enquiries, drafting replies for approval, summarising documents. Visible, cheap errors and a clear before-and-after.
Invest when you have a specific, repeated task involving unstructured information, and the data that task needs already exists and is accessible. Investing because AI is expected of you produces pilots that never reach production.
You have read what we think. If you want to know what it means for your case specifically, describe it and we will tell you which parts of when should a business invest in ai? actually apply — and which do not.
