Financial Modeler's Corner

The episode provides practical, real-time testing of AI financial modeling tools with substantive observations about model structure, formula choices, and scenario management. However, much of the content consists of live walkthrough narration, repetitive affirmations, and meandering discussions about AI fatigue that dilute the insight density. Core learnings exist but are scattered rather than concentrated.

“It basically built three separate models. It didn't know how to run a base best and worst off one model with scenario management or sensitivity tables or switches.”

“This is not how you want to build the revolver section on, uh, so many issues with this. First of all, that should not be right on the financing section of a cash flow statement.”

The episode's core concept - direct AI tool comparison for financial modeling - is sound and somewhat novel. However, the analysis relies heavily on conventional financial modeling best practices (balance sheet balancing, scenario switches, proper depreciation schedules) rather than challenging assumptions or proposing counterintuitive frameworks. The discussion about AI as a learning tool versus automation shortcut is sensible but not particularly original.

“go ahead and build the model in five minutes, get fired in 10.”

“AI Is a magnifier. The better you know what you're doing, the more you can get out of it.”

The three hosts - Ian Schnorr (Financial Modeling Institute Executive Director), Giles Male (MVP, Full Stack Modeler co-founder), and Paul Barnhurst (FP&A professional) - are credible practitioners with demonstrated expertise in financial modeling. However, they function primarily as co-hosts facilitating a tool test rather than as independent guests bringing novel perspectives or outside industry expertise. Their authority is real but the format limits the value extraction from their seniority.

“Giles, male, humble MVP and co founder of Full Stack Modeler”

“Ian Schnorr, Executive director of Financial Modeling Institute”

The episode contains many specific technical observations tied to actual model outputs (e.g., revolver calculations using LET functions, deprecated semi-fixed ranges, hard-coded values in assumption pages). However, these references are anecdotal and tied to this particular test case rather than systematic evidence. No broader data about AI model performance, error rates across models, or quantified improvement metrics are provided. The specificity is granular but narrow in scope.

“That's what it did on your cash. So if you were, uh, going to. Oh my gosh. Holy. That's what it did on your. Wow. This just. Wow. ... really, are you going to use a let function, One of the newer, more complex functions, um, in Excel”

“B41. Let's go to B41. And yeah, so it did a calculation here for net revenue, B38. So units sold times gross price minus freight divided by a thousand.”

The hosts ask reasonable follow-up questions and push back on tool outputs (e.g., questioning the use of LET functions, calling out hard-coded values). However, the conversation frequently meanders into meta-discussion about AI fatigue, learning processes, and philosophical tangents rather than drilling deeper into root causes of model failures or exploring edge cases. The questioning is competent but not sharp or aggressive enough to extract maximum insight.

“But what's it done on the forecast? How is it picking one of the three models with the three ampers?”

“you never go cash positive. You don't have a minimum balance. Yeah. I mean you should be able to. You should be paying it off or. But yeah, get it.”