I have a confession.

Many of you have asked me to run content on AI, and so far I’ve avoided the topic like the plague.

Why?

I have nothing to say. Not a thing.

It’s clear it’s coming. It’s 'gonna be huge'. But when? And what does it mean?

Who knows…

Certainly not me. I’m just a miserable old CFO who’s spent his career keeping an eye on technology but has never been an early adopter.

This wave of AI feels different though. The finance profession is going to change more in the next 5 years than it has in the last 30.

Or maybe not… and it’s just the next Silicon Valley hype train, designed to give tech valuations a boost in the face of the valuation multiple pressures of the last two years.

I had to get closer to the answer…

But I’d need some help. I’ve made four new friends, at the cutting edge of technology and finance.

In addition to your normal Saturday serving, we are running a special series of interviews over the next 4 weeks diving deep into what AI means for finance.

It’s an all-star list, and this week starting with Eric Glyman, CEO of finance automation platform Ramp.

Eric is a founder in a hurry, building Ramp from 0 to a rumored $300m+ ARR in 4 years. At their most recent funding round, Ramp was reportedly valued at $7.65bn. This thing is a rocket ship.

With over $40bn of corporate spend managed on a platform at the bleeding edge of AI in finance, who better to ask what this all means?

Secret CFO (SCFO): “In what ways do you believe AI will change accounting and finance over the next five years?”

Eric Glyman: “The ultimate view I have here is that AI will usher in a new phase of ‘self-driving money.’ What I mean by that is many tasks that need to be done manually right now will happen without you even asking. It will be the ultimate zero touch. So instead of having to “do” certain tasks, finance teams will be in the “verification” and “improvement” business - for example, data entry will happen automatically and with more accuracy. 

There will also be less need to be a technical expert. Natural Language Interfaces will make it possible for users to initiate tasks like creating reports or requesting receipts through simple text or voice commands.

Big picture, I think AI will massively transform how finance teams spend their time. With the routine tasks automated, it gives finance teams room to be more strategic. Over time, I think AI has the potential to end the 50-year productivity slump we’re in right now. That might take more than 5 years, but it could happen in 10.”

SCFO: “You talked about AI's potential to create ‘self-driving money.’ If you could fast-forward five years, what do you hope to see as the biggest impact of AI on the finance industry?”

Eric: “Finance is fundamentally about setting rules in place for how dollars should flow out of (and into) an organization, measuring the impact, and improving return on investments. While there are many rules that drive how and why organizations spend, and how to model these financial decisions, I think that ultimately finance is an exercise in systems design and systems thinking.

As AI’s reasoning capabilities develop quickly, and software is able to “think” it will increasingly be possible for finance leaders to design the key outcomes a company is trying to drive, and set rewards for AI to efficiently achieve these outcomes. More practically speaking, self-driving money and AI’s impact on the finance industry means:

How CFOs can make sure they don’t get left behind…

SCFO: “How should CFOs who know nearly nothing about AI (like me) immerse themselves? How do we make sure we don’t get left behind?!”

Eric: “There are a number of things to watch, read, or listen to - like Intro to Large Language Models, The Dream Machine (great history of the Internet and early AI), Building the AI Platform for Finance Teams, and Ramp and the AI Opportunity.

Also, don't be afraid to experiment. Give your team a budget to test and develop a taste of emerging products. Start small, and most importantly, rather than trying technology and looking for ways to apply it - come up with actual problems in your business and seek solutions that can solve them. We just released our latest quarterly benchmark report, which looks at how 25K businesses are spending with vendors. In the past year, spend on AI tools has 2x’d on cards - which is an indication that more of our customers are experimenting with these tools. Even more interesting, AP spend for AI tools has 3x’d, which tells you that long-term contracts are growing even faster. The best time to start really is now.”

How AI is different this time…

SCFO: “I’ve had machine learning running on Robotic Process Automation (RPA) bots in my AR and AP teams for years. How is this current AI hype cycle different? Or is it just clever marketing out of Silicon Valley designed to pump valuations? Or am I just old and cynical?”

Eric: “I love this question because it gets to the fundamental alignment between AI (the tech) and connected systems (operations). To get the full value, both need to be in place.

To start with the former: what makes LLMs different from the traditional RPA bots is their ability to (1) correctly apply learnings and context from the outside world to an individual company’s data set and (2) accurately predict, generate, and operate when there’s missing information. To put it very simply, it’s just more intelligent.

Looking at the systems piece - to get the best gains in intelligence happening, data within the company needs to be connected and flow together. Otherwise, you’re going to have “intelligent pockets” but not “system intelligence” - and pockets have limited benefits.”

How AI is facilitating/defending against fraud…

SCFO: “One of the things keeping me up at night about AI is payment fraud risk in spend. Where are the emerging risks? What would you recommend CFOs be doing?”

Eric: “To be blunt, I think the main risk for CFOs today isn’t related to new technology, it’s generally related to bad systems, inadequate controls, and cumbersome processes for managing spend.

One of the major emerging risks is around deepfakes. Gen AI makes it possible to generate deepfakes in images (like invoices), video/audio (impersonating a vendor calling you), and generally trying to get employees to break a process. Things like this are getting good fast, and fraudsters are using them today.

This is where having a standard system (Ramp being an example) can help - because the platform can detect things like doctored images in ways humans might not be able to. That said, if you want an interesting site that articulates these types of fraud risk well, I suggest Reality Defender.

On the flip side, AI has the potential to manage risks in ways not possible a few years before. For example - this story about a former Yale employee who stole $40M in electronics. That’s a simple use case for AI.

SCFO: “Could you give a quick rundown of the main products/features Ramp offers?”

Eric: “Ramp is a command and control system for company spending, with a suite of products designed to streamline and automate financial operations. On Ramp, you can: 

Automation (AI) is incorporated in all aspects of the platform - the goal is to be as zero-touch as possible. For example, we’ll detect any overspending or duplicative vendor spend and automatically flag that for you.

SCFO: “But P-Cards and corporate cards have been around since god was a boy. How is Ramp different?”

Eric: “Ramp cards are programmable and can be created to only work at specified merchants, expire after as many or few uses as you want, and can automate the collection of receipts/invoices. Classic P-cards lack these controls and, worse, lack automation that connect to the broader workflows.”

What this means is that:

Besides being beneficial for cost, customers told us this was an emotional relief too - not having to play bad cop and nag, telling an employee they can’t spend on x item or that they need to submit 30 missing receipts, and then later running into that person at the water cooler… not an enjoyable experience.”

SCFO: “How does Ramp interact with ERP systems? Both modern ERPs and legacies, if there is a difference?”

Eric: “The value we bring to customers is not just the tech we create ourselves, but how we help our customers connect their own systems internally and achieve a continuous close. Ramp integrates with a number of leading ERPs.

I can’t name the specific person, but I was meeting with the CFO of a major social network last week, and he was showing me all the systems (100+) they use for finance (across AR, AP, ERP, etc.). It literally took up half a wall! What we’re doing is breaking those silos and points of friction down.”

SCFO: “Managing the long tail of spend is a consistent headache for CFOs. It’s an incredibly high volume of low-value, disparate transactions. So it’s hard to ‘procure’ it. Stationery, engineer call-outs, team pizza, replacement door handles, reception desk flowers... I could go on. Individually irrelevant but significant in aggregate. It seems like Ramp is best suited to that ‘long tail’ spend, rather than say, COGS spend. Is that right?”

Eric: “Long tail spend was how we started - with one unified platform to manage cards and expenses - but we’ve expanded significantly since then. Now we can handle all costs associated with running a business except for payroll.

We think there’s a massive opportunity to create an intelligent operations platform for finance that links all spend - including COGS - together.

SCFO: “You are handling $40bn of corporate spend every year. What is the one thing that CFOs typically are not doing, that they should be?”

Eric: “We believe too few CFOs are thinking about improving the architecture of the financial and operational systems they use. A big reason they aren’t doing this is because most of the systems and tools today aren’t good and so they’re just making do with what’s available. We’re obviously working to change that. I believe the next 5 years we’ll have software that both thinks and acts on your behalf. That will bring immense opportunity for CFOs to rethink not just how their teams run, but how their whole company operates.”

Ramp’s business model…

SCFO: “And how does Ramp make its money? Interchange + SaaS revenue. Anything else? Are there any supplier kickbacks?”

Eric: “Ramp primarily earns money from Interchange (we keep a small portion above the cashback we share with customers), SaaS revenue from customers that upgrade to Ramp Plus, foreign exchange from international money movement, float on bill payments in flight, and affiliate fees when flights or hotels are booked through Ramp Travel. What makes Ramp so powerful for customers is that between the hard dollar and time savings we drive for customers, and the cashback we rebate, the ROI of using Ramp shows up clearly and quickly on companies’ bottom lines.”

On using AI internally…

SCFO: “I read you are using AI-led tools internally to summarize customer calls into a ‘feedback podcast’ for the team. Do you have any other examples of ways you are using AI internally?”

Eric: “Yes, many. Internal efficiency is a big focus for us and in many ways is a reason why we’ve scaled so quickly with a relatively small team. A couple examples…

Paired engineers with sales reps to create automated systems for outbound motions - helps with research, email drafting, etc. We estimate that this has made our average SDR 3-4x more productive than our next closest set of competitors. We get deeper into this here.

Toby, our “AI teammate” who listens to recorded customer calls (100K+) and answers questions from our team via Slack. Questions as broad as why someone might choose Ramp over a competitor to something more nuanced like what features customers in the construction industry are asking for. Here’s what that looks like…

One of the things on my mind right now is creating a cultural norm around “full stack builders.” How can we make sure every person at the company is able to leverage systems and Large Language Models to augment and do work for them?

I believe in disciplined operations. We’re doing this for our customers by making it easy for them to save time and money and redeploy that in their companies. And we’re doing the same for ourselves.”

SCFO: “You started your career in financial restructuring, and now you are running a unicorn disrupting corporate spend management. What on earth happened in between?”

Eric: “I’ve always been interested in how great leaders run their businesses. And one clear theme is they’re all disciplined operators and manage their costs like a hawk. They want to know how they can reinvest every dollar and every hour back into their businesses.

My role in financial restructuring showed me that even once great businesses - if they did not apply rigor to how their teams spent time and money - could go bankrupt. It also taught me to look for value that people missed or overlooked. 

It was in this role that the idea of my and Karim’s first company, Paribus, formed. Paribus was a consumer spend management platform that was effectively an AI agent for refunds on retail items. 

Fast-forward a year, and we had almost 1 million users and were acquired by Capital One - becoming what’s today Capital One Shopping. We spent a few great years there, but it became clear in that time that there was an even bigger opportunity than Paribus had. Karim and I knew we had the opportunity to start a new kind of card company - one that focused on helping customers spend less, not more - in a market that had been stagnating for 50 years. That was the start of Ramp, and our platform and vision have dramatically expanded since then.”

SCFO: “Looking back at your journey from financial restructuring to leading a unicorn, what was the most unexpected lesson you learned along the way?”

Eric: “The more time in restructuring I spent, the more I grew to like Tolstoy’s opening line in Anna Karenina: ‘All happy families are alike; each unhappy family is unhappy in its own way.’”

There are many, many reasons that a company can fail and end up in bankruptcy – liquidity issues, solvency issues, business disputes, etc. They get unfocused. Somewhere along the way, they lose the strongly unifying magic needed to keep an operation running. Complexity kills.”

SCFO: “What was one of the most challenging moments in your entrepreneurial journey, and how did you overcome it?”

Eric: “It’s interesting because one of the most challenging moments was actually the same for both Paribus and Ramp – the challenge of being misunderstood and dismissed early on. Getting told “no” a lot. I’ve gotten to know many entrepreneurs and investors over the years and it’s the biggest wall many face out the gate. If you’re doing something truly contrarian you can almost guarantee it’s the first, consistent, and most challenging moment you’ll face. It kills most ideas before they can blossom into something bigger. You could call getting through it grit or determination. But now that I’ve experienced it a couple times I also know if you don’t get this reaction in the beginning, chances are what you’re doing isn’t as contrarian as you might think.”

Reflecting on this interview made me think about this quote from Luca Maestri, Apple’s CFO who stepped down last week after 11 years:

He is running investor relations for a business with more than $3 trillion market cap with just 2 people.

Good people will focus on the important not the urgent, automate as much as possible, make their jobs simpler and end up 10x more productive per hour.

That includes finance. And especially when they are powered by AI. Ramp is on a mission to make that easier for CFOs. It’s worth checking them out.

A huge thank you to Eric and the Ramp team for sponsoring this series, and making time for the interview. We’ll be back next week for the second of four ‘One-on-one’ interviews diving deep into the future of AI in finance. 

And, as always, you can expect your regularly scheduled email on Saturday, with a new series starting on investor relations.