For Episode 27 (January 16, 2026), Sam built a short quiz with our guest's company, Espresso AI, in mind. It asks a data team five questions about how they use their warehouse, then returns a "CFO BPM" score, an estimated monthly saving and key observations. Sam didn't say how long it took or what he built it with, so this page focuses on the quiz itself and the reaction.
Espresso AI uses machine learning to cut Snowflake and Databricks SQL compute costs in real time. Sam wanted a quick, playful way for a potential customer to see how much they might be wasting on their data warehouse before talking to anyone.
Not covered on the show. The app has three parts: a company-size picker, a five-question questionnaire, and a results screen with a CFO BPM gauge, an estimated savings figure and a list of observations.
The $8,000 is the quiz's own estimate for a made-up team, not a measured result.
Espresso AI CEO Ben Lerner was clapping off-mic. He called it "a super, super fun quiz" and the kind of thing people should think about. Then came the twist: the questions point to FinOps and budgeting discipline, like noticing you haven't watched Netflix in six months and canceling it. That's "exactly the opposite" of what Espresso does. It plugs into the warehouse, isn't opinionated about what you run, and optimizes in the background.
But you can also not do a bunch of work, and we can lower your bills for you using AI magic.
— Ben Lerner, Episode 27
His read on the result was practical: if the bill is that high, it's probably a team with two data engineers who already have too much on their plate. He'd rather hand them a button that cuts the bill in half than a list of chores.
Frame the results around both paths. Ben's view is that manual cleanup is worth doing and becomes necessary at scale, but automated optimization can lower bills without the team touching anything. A better quiz would say which problems need discipline and which can be automated. Jordan related to the pain: at 05:14 he mentioned a GPU cluster we left running and an AWS overage, and wished we'd had guardrails in place.
The rest of the episode covers why Ben thinks AI isn't yet good enough for serious backend infrastructure. More in AI infrastructure and AI in developer tools.
Sam's quiz asks five questions: how often you use SELECT * in production, whether you know your top five most expensive queries or dashboards, how often dashboards refresh versus how often they're viewed, whether multiple teams query the same raw data, and who owns warehouse cost.
Our version is a short questionnaire that returns a 'CFO BPM' gauge, an estimated monthly saving and key observations. The savings figure in the demo ($8,000 a month) came from the quiz's own logic, not a real bill.
Not necessarily, according to Espresso AI CEO Ben Lerner. FinOps discipline helps and is needed at scale, but Espresso plugs into the warehouse and optimizes in the background, so teams don't have to do the work themselves.
