What would count as a useful AI result?
Define AI success around the whole business task: a baseline, useful quality measures, the cost of errors, and a decision about what happens next.
Read articleHalsted, Haines & Co. / Journal
Notes on better decisions.
Original perspectives on strategy, research, and applied AI from Halsted, Haines & Co., a State Madison company.
When a business question needs research before AI
A practical way to distinguish an information task from an unresolved business decision before designing an AI system.
Read article
Who owns an AI system after it starts being used?
Clarify the decisions, information, technical responsibilities, and escalation routes that continue after an AI system becomes part of everyday operations.
Read article
Choose a workflow before choosing an AI tool
Compare possible AI uses by the task, available information, review effort, dependencies and the business decision an improvement would support.
Read article