FinOpsPod

The episode contains a genuinely useful methodology for LLM implementation (exhaustive data pre-processing before any AI involvement, building MVP without LLM first) that goes beyond typical AI hype, but roughly 8-10 minutes of the runtime is intro banter, outro, and host summaries that dilute the idea-per-minute ratio.

“if you want to build something with Genai or large language models, just build it and have your MVP be completely without a large Language model. If you can do it without, you're good to go. Then you can add an LLM on top”

“the LLM is a tiny fraction. If it was a Python script with 20 functions it would be like one of the 20 functions. And that's it.”

The inversion of conventional AI-first thinking - pre-processing data into near-complete analytical sentences in Python before touching an LLM - is a fresh and counterintuitive framing; the sustainability-via-LLM tangibility argument (trees, light bulbs) is genuinely novel for FinOps context, though neither idea is paradigm-shattering.

“You want anything that does an analysis to have the potential to disagree with you and to find things that you cannot, because those would be useless analysis.”

“we can assume that if it has a higher dollar cost, it also has a higher carbon emission. That way we can create a report where we actively, without very much, very little effort, say this is your dollar cost”

Benjamin is a genuine hands-on practitioner who has built and deployed these systems at scale at what is clearly a large enterprise (J&J implied), presented at FinOps X Europe, and received a promotion - grounding his claims in real production experience rather than thought-leadership abstraction.

“I'm one of the lucky people to have a large finls team around me. The fact that we now don't have to spend so much time creating these individual analyses, these reports, these deep dives, because we can within eight seconds have a root cause analysis”

“we have all this cloud billing and usage data, literally hundreds of millions of lines per day”

The episode offers some concrete specifics - seven-to-eight second processing times, hundreds of millions of billing lines per day, networking inefficiency examples like net gateway vs endpoint - but lacks hard metrics on cost savings, team headcount, error rates, or dollar figures that would make the claims fully verifiable.

“your traffic for example, from DCC2 to S3 is going overnight. That's the recommendation we give. It's a typical inefficiency where you push your traffic over a net gateway instead of an endpoint”

“scope three, which is 40 to 50 to even more percent of your emissions”

The host does solid synthesis work, connecting the guest's framework to a prior episode and surfacing the LLM-sycophancy point effectively, but never genuinely challenges claims, pushes for harder evidence, or probes failure modes beyond what the guest volunteers.

“LLMs will always agree with you. So that's a danger.”

“You're still doing all the work. What you've done is you are defining what is good and what is not good.”