Fixing AI Budgets: Tying Spend to Customer Benefit

Companies are burning through entire year's AI budgets in months... and admitting they can't tie the spend to any real customer benefit. My latest for Forbes on fixing this: → 𝗧𝗿𝗲𝗮𝘁 𝗔𝗜 𝗯𝘂𝗱𝗴𝗲𝘁𝘀 𝗹𝗶𝗸𝗲 𝗵𝗲𝗮𝗱𝗰𝗼𝘂𝗻𝘁: just like you wouldn't give any team an unlimited headcount budget, no team should get an unlimited token budget. → Plan quarterly, tie every dollar to an outcome. → Teams that are outperforming on driving outcomes should earn more AI budget. → Build cost discipline into the system and a culture of cost consciousness. → Kill the public token leaderboard. It rewards burning tokens, not building great products. Full piece linked in comments 👇

  • graphical user interface

Insightful piece Vinay - particularly the observation that the winning teams are starved and the losers are hidden Interested to hear what how you guys are measuring outcomes and expectations of what the lag should be between token consumption and outcome improvements

Great read. I, too, am for killing token leaderboards which removes a bad incentive, but I also believe there should be a leaderboard to promote teams that are leading outcomes with AI so other teams get visibility into what’s working well.

That pattern of rapid spending without a measurable customer outcome suggests the initial business case was built on hype rather than a specific problem to solve. I've seen teams get better results by starting with one clear operational pain point and scaling only after they have hard evidence it's working.

The headcount analogy sounds intuitive, but it ignores a critical difference: adding one more person is a step-function cost with onboarding lag, while spinning up extra inference tokens during a viral demand spike can be pure margin if the product monetises that usage. Treating AI budgets like fixed headcount could leave a company unable to capture windfall opportunities. Maybe the real fix is dynamic, usage-based budgets that flex with proven value signals, not a quarterly personnel-style cap.

Like
Reply

That admission is actually kind of refreshing in a strange way because so many teams are too deep in the rollout to stop and ask what changed for the customer. The hard truth is that a useful pilot costs almost nothing, and scaling something nobody truly wants burns cash fast.

Well said, Vinay Kuruvila . This is how budgeting should be treated and challenged. It will enable teams to plan more effectively rather than assuming that, once requirements are provided, everything can be built in no time. Impact and value should play a much bigger role than simply checking boxes on leadership boards.

Great piece with real world experiences!

You guys definitely don't make it easy to get in touch with. But I am committed to getting resolution by way of refund for your non-working dating apps.

Like
Reply
See more comments

To view or add a comment, sign in

Explore content categories