How can AI-based decision models improve my business strategy?
AI-based decision models help leaders move from intuition to evidence. The question is how much they actually help, and that has been measured.
Dr. Deshpande's doctoral research at Singapore Management University tested this. Senior executives were asked to judge real market-entry decisions whose outcomes were already known. Working from experience alone, they were right 68% of the time. An AI model working alone reached 70%. Executives working with the model reached 77.3%, a 14% improvement over the baseline, statistically significant at 95% confidence.
Neither humans nor AI won on their own. The combination did.
One finding shaped how these engagements are built. When an untrained AI was given the same decisions, it also scored 70%, but every error it made was a false "go," each resting on a confident assumption that conditions would hold. In market entry, a false go is the expensive error. So the model ranks the evidence and exposes the risk; the decision stays with you.
Who it's for
Boards, executive teams and advisers weighing strategic options like market entry, expansion, or major reallocation. The goal is to reduce bias and decision fatigue without handing over the judgment. The approach works best for leaders who already know their business deeply; it sharpens expertise instead of replacing it.
What you get
A structured engagement that analyses your data, builds a decision model for your business, and delivers ranked recommendations with the evidence and boundary conditions made explicit. That includes what would have to change for the recommendation to flip. Models are revalidated as the market moves; a framework built for one sector is retested before it is applied to another.Research background: Deshpande, T. (2024). Doctor of Business Administration, Lee Kong Chian School of Business, Singapore Management University. Examined by a committee spanning Computer Science, Operations Management and Strategic Management.Published and openly archived →