From Spreadsheets to AI Agents
Recorded live at the Gurobi Summit in Las Vegas, this episode follows optimization from the spreadsheet to the AI agent. Two partners who worked with Gurobi’s founders before Gurobi existed, explain how optimization moved from the Excel Solver to AI agents, and why many more business problems are now worth solving.
Daniel Fylstra, founder and president of Frontline Systems, built the Solver that ships in every copy of Microsoft Excel.
Christian Valente, senior software engineer at AMPL, works on the modeling language that has connected to Gurobi since version 1.0 in 2009.
The conversation spans the origins of the Excel Solver and the AMPL modeling language, Frontline’s early role as a Gurobi reseller, and the growth of spreadsheet optimization from 200 decision variables to models with 2 million. Daniel and Christian also discuss resource allocation across hospitals, refineries, and sawmills, explainable optimization in gas pipeline scheduling, and how AI agents and LLMs are changing who can build a model.
Central to the discussion is a clear thesis: as AI lowers the cost of formulating models, companies can solve far more of their problems, including those worth $20,000 to $50,000, but people still need to supply the judgment and keep optimization transparent to the teams who rely on it.
In this episode
Why a hospital, a refinery, and a sawmill solve the same kind of resource problem.
Explainability as a business requirement in gas pipeline scheduling.
How AI agents turn a plain language description into a cutting stock model.
What skills and MCP servers mean for leaders adopting AI.
