The Token Budget Problem Nobody Is Talking About

Giving every employee the same AI allowance can miss where additional spending would produce useful work. In this interview, the chief executive of Factory, which builds software-development agents, argues for allocating budgets according to the benefit of the next increment of usage. Different departments may need more model usage, more people or better data. His proposal makes AI budgeting a question of which work improves when spending increases. He expects companies to develop better ways to measure that benefit; the interview does not present a validated allocation method.

“The future of software engineering is, you're not building the software, you're building the factory that builds the software.”

Matan Grimberg

“The biggest threat to OpenAI and Thropic or any of the model apps is not each other, it's open models.”

Matan Grimberg

“There is no way in 12 months, orgs are going to do such a kind of blind allocation of tokens where everyone gets the same.”

Matan Grimberg

“No one gives a shit if you are early, you don't get a consolation prize.”

Matan Grimberg

“I think if we succeed, our customers will know the model that they're using just as well as they know, like where the electricity that powers their toaster comes from.”

Matan Grimberg