When I first became a finance leader, I often felt like monetization was a black box, like something product owned until finance needed a forecast. With AI, that tension only gets sharper. That’s why I’m excited about this free webinar with Martin Casado, General Partner at Andreessen Horowitz. On November 4th, he’ll share what he’s seeing across the market as companies experiment with new monetization models. Learn how leading startups and public companies alike are adapting pricing, packaging, and governance for the AI era.

This post begins a four week march to our most anticipated report of the year: The State of the Agentic Financial Stack (I’ve told all my neighbors about it, even the one who’s a firefighter and doesn’t understand how the guy down the street sends emails for a living).

For the last three months I’ve been in the lab.

I’ve spoken to founders and execs who are supercharging the tools we love and use each day with AI - from ERPs to Close Management to Invoicing and Billing to Procurement - to figure out wtf is actually real and game changing.

The output is a comprehensive report, with market maps built for each distinct layer of the cfo tech stack.

We. Love. Market. Maps.

It’s like Gartner, without the librarian tone or the software-like price tag (my report is free AF)

I’m sending the report early to a select group of finance leaders. Sign up to be one of them.

Instead of telling you what AI can do for finance (when they zig, I zag!), this week I'm going to tell you when AI is complete overkill for your finance stack.

Consider this your bullshit detector for the next three vendor pitches that land in your inbox.

Every sales pitch is using "AI-powered" as a modifier on slide 3 (and 4, and 6, and 9). Every LinkedIn post promises agents will revolutionize your close process (also, WTF is the correct definition of ‘an agent’???). Every conference panel asks the same tired question: "How will AI transform the office of the CFO?" (but also, please come listen to me speak, I have day care bills).

Here's what no one's saying:

I've been deep in research for an upcoming report on agentic finance tools, and after dozens of conversations with founders and CFOs actually building and buying this stuff, I'm convinced we're at peak AI hype in finance.

So let's talk about what the vendors won't tell you.

(And look, I love a good buzzword as much as the next guy… I’ve been using the word “exogenous” on my podcast a lot... But ya boy also HATES watching CFOs get sold snake oil when they should be, you know, running their business.)

Before evaluating any AI tool, ask yourself:

If the answer is yes, agents are overkill. Full stop.

That's it. That's the tweet. You can close this email now if you want.

...still here? Cool. Let's keep going.

Goodwill impairment tops this list.

Goodwill impairment isn't a calculation; it's a series of judgment calls stacked on top of each other.

What's the right discount rate for a reporting unit? Which market comps are actually relevant? How do we weight qualitative factors like management changes or competitive threats? What growth assumptions are reasonable given current market conditions?

Each of these requires understanding your business strategy, industry dynamics, and the specific circumstances, especially for an acquisition.

AI can pull comps and run sensitivity analyses, but it can't make the call on whether your reporting unit's fair value dropped below carrying value because of a crappy quarter or a fundamental shift in the business.

Other no-go zones:

My standard: When dealing with money, AI needs to be better than a human, not "almost as good."

If you are changing my dinner reservation time, you can be almost as good. If you are paying for my flight, you need to be better than a human (see: Air Canada).

Don't pay AI prices for workflow automation.

(I mean, unless you're into lighting money on fire. In which case, call me; I have some premium mostly metrics branded matches to sell you.)

Your ERP already handles:

Your existing expense tools already solve:

One of my favorite quotes came from a CFO I spoke to as part of the research:

To underscore his point: AI needs to be additive, not duplicative. If you already have an ERP and a planning tool (shoutout Abacum) that does something well, you don't need AI to replicate it. Yes, you can use some AI features to super charge it. But don’t go looking for a problem that already has a solution. That’s a very expensive side quest.

Some things legitimately require human judgment. That's not a technology gap; it's a premium feature. You should be willing to pay more for it in the form of skilled labor.

AI can't fix messy upstream data; it just amplifies the mess.

Or as a CRO I used to work with used to tell me:

If you have:

You don't have an AI problem. You have a data problem. Fix that first.

One CFO I talked to put it more eloquently than my CRO friend:

You can't sprinkle AI pixie dust on garbage data and expect gold (you get expensive, hot garbage.)

We're seeing a lot of rebranded chatbots being sold as "agents."

(Also, Drift was really before it’s time. Today it would be rebranded as an AI chat bot and sell for 100x ARR.)

First off - what is an agent? My simple definition is an agent is software that doesn’t just respond, it owns outcomes. It can deal with fuzzy logic and go from point A to Z (with stops to gather intel at spots L, M, N, O, and P along the way). That’s the difference between traditional automation. It’s goal seeking.

What’s not an agent (or at least not a useful one?):

I’d watch for these red flags:

Keep your vendors honest. From speaking with Numeric’s Co Founder Anthony Alvernaz:

If they can't answer clearly, GTFO.

So to drive the point home, here are some areas where AI is actually too much.

The AI gold rush in finance has created a paradox: vendors are trying to apply agents everywhere, which means they're often solving problems that either:

The best CFOs I know aren't asking "How can I use AI?"

They're asking: "What problems do I actually have that AI uniquely solves?"

That's a much shorter list than the vendor pitch decks suggest.

Next week, I'll cover the flip side: the "agentic sweet spot" where AI genuinely delivers value that traditional automation can't match.

Spoiler alert: it's the messy middle… too much variability for RPA, but enough structure to learn patterns. Think collections, pipeline forecasting, procurement end-to-end.

This is part of my research for "The State of the Agentic Financial Stack" report, launching in November. We're talking market maps, case studies, real metrics from real deployments, and frameworks for actually evaluating this stuff.

If you're building in this space or adopting these tools, I'd love to hear from you.

Sign up here to get the full report when it drops (it's free, and I spent hours picking out the memes)

In this episode of Run the Numbers, I sat down with my buddy Ivan Makarov, Operating Partner at Andreessen Horowitz and former VP of Finance at Webflow. Consider this your roadmap for building finance teams from scratch inside hyper growth companies. We discuss deciding between outsourcing and hiring in-house, what makes a great first finance hire, and why early-stage companies often run out of cash before they run out of ideas.

Ivan also dives into fundraising pitfalls, audit readiness, and the tools that make up a modern finance stack. Beyond the spreadsheets, Ivan opens up about the shift from operator to venture partner, the value of helping founders avoid his past mistakes, and what makes an offsite actually work.

Did you know the US Postal Service has a podcast now? Kyle Poyar and I cover this super exciting news, plus the circular economic doom loop we find ourselves in, with AI propping up our economy.

We break down the hierarchy of business model monetization, and if ChatGPT moving into Slop Tock and Erotica signals a bubble is about to burst.

Most operators bring in a lawyer after they’ve hired the banker. On paper, that seems efficient: banker kicks off the process, lawyer handles the docs. Church, and state.

But in practice, you’ve already given up leverage.

I’ve made this mistake. So have other smart operators I know.

You sign an engagement letter, get a banker running, and by the time legal gets involved, the structure is already locked. Terms that sounded harmless at the start… fees, tails, carve-outs… turn into multi million dollar surprises later.

I spoke with two experienced M&A lawyers on my podcast, Trey O’Callaghan (Goodwin Procter) and David Siegel (Grellas Shah). They’ve advised on hundreds of tech and PE-backed deals. Their advice was consistent and clear:

Here’s what you’re missing.

Hoping we aren’t in a bubble,

CJ