I Want My Team Asking for More AI Capacity

I want my team pushing frontier AI tools to their full potential on valuable work. If the allowance gets in the way, a request for more capacity is a useful signal. I have no interest in a dashboard ranking token consumption or encouraging people to maximize it. Success means the work gets better.

Using AI to help write evaluation tools to measure the quality of the latest algorithms is one example. There is room to use AI throughout that process, from developing the tool to checking whether the evaluation itself is reliable. I want to know whether that lets us examine algorithm quality more thoroughly, reach a useful evaluation sooner, or reduce the effort involved.

Those questions have to include the work after generation. A tool that takes less time to write can still require substantial inspection and correction. I care about how long it takes to get to an evaluation we can trust and use, including the human effort needed to get it there. Better quality may justify spending more time. The benefit needs to be visible in the work, even when it does not fit neatly into a claim about hours saved.

Running into an allowance tells me demand has reached the capacity we provided. A small allocation can be exhausted quickly; a generous one may support substantial use without anyone asking for more. Repeated attempts to fix an unreliable result can also consume a lot of capacity. I want to understand those differences before drawing a conclusion about adoption or effectiveness.

I still want people asking. When someone can explain the work, show what is improving, and identify what additional capacity would enable, I want us to take that request seriously. There also needs to be room to experiment before the benefit is certain. Requiring proof of savings before anyone can explore a new workflow would make it difficult to discover what the tools are useful for.

For large corporations, I think this will become a recurring budgeting problem. As AI becomes part of a product manager’s daily workflow, access to a tool and enough capacity to use it become separate decisions. The work may change faster than an annual allocation anticipates. Finance needs visibility into spending, and the team needs a way to expand useful work without repeatedly stopping for approval. Different tasks will need different amounts of capacity, so giving everyone the same allowance will not settle the question.

I want an AI budget that can grow with the work. If a team is getting better at evaluating algorithms, I want to understand what made that possible and whether more capacity would help. If spending rises and the work is not improving, that deserves scrutiny too. I want enough visibility into the work to decide when to expand the budget and when to question how we are using it.

This writing reflects my personal perspectives on product management, AI, and content discovery. It does not represent the official position of my employer or any affiliated organization.