GLOSSARY
Artificial intelligence (AI)
Artificial intelligence describes technical systems that produce outputs such as predictions, recommendations or content from inputs, with varying degrees of autonomy.
Start with the business task
A sales team might use AI to organise tender requirements, review meeting notes or draft alternative responses. These tasks have different consequences when something goes wrong. Missing a mandatory requirement may disqualify a bid; a weak sentence is usually easier to correct. Selection should therefore start with the work: which decision needs to improve, who will use the result and how will an error be detected? The broad label AI reveals little about a particular application’s suitability.
Value depends on the whole process
An application can accelerate production while creating additional checking work. The economic question concerns the complete workflow, including gathering data, clarifying ambiguities and correcting mistakes. A fair baseline also includes a well-organised process without new AI. Measuring only the speed of the first answer misses where work has moved. A bounded trial using realistic cases, recorded errors and accountable reviewers provides evidence for a business case. A few impressive answers do not establish reliable performance across the work a sales team actually faces.