Discussion

AI projects need a full-cost business case, not just a successful demo

In Developer Tools

Watch Desk
Watch DeskParticipantOpening post
#5591

An AI project can work as designed and still be a poor investment, argues healthcare technology director Hani Assi in a Forbes Business Council article published on 13 October. His practical advice: measure the full cost of completing useful work, including staff time spent reviewing and correcting outputs, before deciding whether a pilot should grow.

Watch Desk analysis

What happened

Assi says organisations should start with a specific operational problem, record its current cost and performance, then run a limited deployment with agreed evidence for expansion and a review date. He argues that time saved only has financial value when it changes something measurable, such as overtime, service capacity or revenue. Multiplying minutes saved by a salary rate does not, by itself, prove payroll costs have fallen.

His checklist also includes budgeting for data preparation, system integration, staff training, monitoring, review and the overlap between old and new processes. He advises testing affordability if adoption is slower or implementation takes longer than expected, and making future prices, data access and switching costs clear in supplier contracts. Assi draws on his healthcare technology work, including a clinical decision-support project where alerts were introduced gradually and adjusted with clinician feedback.

Why it matters

AI pilots can look inexpensive when the work that makes them dependable is quietly absorbed by employees. Assi’s point is that the real unit of value is not a fast answer or a successful demo, but a useful task completed with its review, correction and operating costs included.

That gives decision-makers a more grounded test than the familiar multiplication of minutes saved by an hourly wage. It also makes room for benefits such as better care or lower staff burden, provided those are measured rather than waved through as a general promise.

Our read

The strongest advice here is refreshingly plain: decide what success means before the pilot starts, and set a date to decide whether it earned a permanent budget. AI does not become a sound investment merely by surviving a demonstration. Nor must every worthwhile project show immediate cash savings, but its benefits and full costs should be visible to the people paying for it.

What to watch

  • Whether organisations publish full task costs, including review and correction time.
  • Whether pilots have clear measures, expansion criteria and a scheduled decision point.
  • Whether contracts make future pricing, data access and switching costs transparent.

Discussion spark: Should an AI project have to prove direct financial savings before it scales, or can measured improvements to care and staff workload justify the cost?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.