Future Finance

There is a useful breakdown of the five-layer AI tech stack and the rationale for structured, pre-vetted questioning over open-ended prompts, but these insights are buried under extensive off-topic banter about beards, roulette, in-laws, and tattoos. The core claims - time savings and hallucination avoidance - are stated but not explored in depth.

“rather than just asking AI, hey, what's up? We'll ask it an entirely vetted, targeted list of thousands of questions that rely on domain expertise”

“In a lot of companies there people are just too busy managing the day to day. So there's not somebody who's stepping back often and just thinking about the numbers and looking through the data and doing analyses”

The five-layer pipeline framing (ingest, compress, tag, query, present) offers a marginally structured way to think about AI deployment in FP&A, but it rehashes well-circulated points about hallucination from large datasets and the need to chunk data. No genuinely contrarian or first-principles arguments are advanced.

“if you let agents run, they basically make stuff up or head off in the wrong direction or my favorite thing, sometimes they miss the forest for the trees and sometimes they miss the trees for the forest”

“one of the biggest areas for value creation for them was actually just getting their accounting right because they couldn't figure out their P and L correctly”

Both guests carry meaningful practitioner credentials - McKinsey/Bain Capital/CEO/CFO track for Nick and Paloma Partners/Zenbanto ($1B+ volume) for Dan - but Eagle Rock CFO is brand new and the episode draws almost entirely on prior career stories rather than hard-won lessons from running the current venture at scale.

“Nick is a Harvard MBA out of McKinsey and Bain Capital who then went the operator route. He was CEO at IdeaScale, a B2B SaaS company where he nearly quadrupled EBITDA margins at 18 months”

“Dan is a Stanford chemical engineering grad and also a Harvard MBA who spent almost seven years at Prime Camp Management, which for those who don't know, it's one of the most respected and secretive fund shops in the business”

A handful of concrete figures appear - 20-50 hours/month savings, 1-in-100 AI error rate, $5-6M revenue threshold - but these are unverified self-assessments with no named clients, no case study outcomes, and no third-party validation. Dollar metrics and timelines from the actual product are conspicuously absent.

“I believe our tool cuts that 20 to 50 hour exercise down to a few minutes, uh, of human time tops”

“Nick has told me that, uh, maybe one out of 100 questions the AI will still get wrong”

The hosts repeatedly allow the conversation to derail into jokes about beards, tattoos, in-law visits, and roulette, and the back half of the episode is consumed by AI-generated personal trivia questions. There is no pushback on any product claims, no follow-up probing the error rate or the time-savings figure, and no productive disagreement.

“I go, I've been told by people I need to start a beard brand”

“No neck tattoo is”