Free with a sign-in.
The best 9 min of 29
Ramp Co-Founder and Co-CEO Eric Glyman on AI Adoption, Jobs, and Token Spend
Eric Glyman says Ramp’s spending data shows AI adoption is accelerating—and the companies using it most are hiring more, not less.
IBInside Blackstone · Christine Anderson
- Eric Glyman is cofounder and co-CEO of Ramp.
The best minutes
- 10:30AI-intensive companies are growing and hiring faster5 min
- 15:42Token spend needs its own controls3 min
Also worth watching
- 22:08Ramp hires for standout ability rather than a full checklist3 min
- 5:38Ramp tied its card business to customer savings3 min
- 8:07Ramp tracks organizational friction and resets its meetings2 min
Skip the six-minute opening on market conditions and Ramp's product, plus the three-minute debrief on AI adoption and brand.
Full analysis · 6 min read
Real Alpha · Nov 9
Eric Glyman says $30-per-employee AI spenders hired 12% more
- Glyman put the high-intensity line at $30 per employee monthly.
- Those firms hired 12% more than nonadopters or later adopters, but causation remains unproven.
- Track task-level costs, output, and staffing before treating AI use as a labor substitute.
Free with an account
Read the rest of the analysis
6 sections, the 4 other voices it tests Glyman against, and what it means for you. Every episode we cut gets one.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Signed up before? The same buttons sign you in. The episode, its moments and the transcript stay free.
High-intensity AI spenders hired more, but causation remains open
The clearest number is Glyman’s high-intensity line, $30 per employee per month. He said those firms hired 12% more people than firms that had not adopted AI or would adopt later.
He described that level as roughly a license or two, not a universal threshold. Ramp’s measure captures paid spend, while Revelio Labs pairs jobs data with LinkedIn and other sources.
Glyman said the firms hired more despite growing at the same rates. That is his description of the comparison, not enough detail to rule out differences in industry, size, or demand.
Ramp sells finance automation, and Glyman presented analysis from its customer spending data. The episode does not supply the hiring sample size, study window, control variables, or full matching method.
So the 12% is a descriptive association, not a causal estimate. Founders should treat it as a reason to inspect their own cohort data, not as a headcount forecast.
Our read
The 12% finding pushes back on automatic job-loss assumptions, but it does not show AI created those jobs. Treat the high-intensity line as a prompt for local analysis, not a staffing rule.
Ramp’s highest AI-spend group roughly doubled revenue
At the time, about 50,000 businesses were in Ramp’s dataset, Glyman said. He compared the top quartile of most intense AI spend with businesses that did not adopt.
The period was three years after ChatGPT appeared. His top-spending group, about 12,500 businesses, roughly doubled revenue.
The nonadopting group was roughly flat, though he said it grew about 10%. For example, he named a Utah roofer and a Florida construction firm.
He used those examples to show the top group included firms beyond technology. That makes the observation broader than software, as presented, but does not establish what caused revenue growth.
Fast-growing businesses may have had more room to buy AI, and the episode gives no matched sector or baseline comparison. Nor does a multi-year revenue split identify which AI task produced growth.
The useful question is where a comparable workflow changes output, not whether high spend predicts success everywhere.
Our read
The data show an association between intense AI spending and growth, not proof of an AI growth effect. Use the split to form a testable hypothesis, not to set a revenue target.
Ramp’s payment data suggests adoption exceeds surveys
Payment data can surface adoption a survey may miss, but the measures differ. Glyman recalled what he thought was the latest Census estimate, roughly 22% to 25% of businesses spending on AI.
He said Ramp’s actual spending data pointed to a significantly higher share. He later said he thought about 55% of businesses in its dataset already spent on AI.
For companies in the dataset, he put median monthly AI spend at about $12 per employee. For the top one percent, he said it was almost $10,000 per employee per month.
That spread matters more than a single adoption headline. A company with a small AI charge may be experimenting, while a top spender may have a very different use pattern and budget.
Glyman said the Federal Reserve’s FRED database began drawing on Ramp data directly as a more real-time signal of business spending. That supports timeliness, not an audit of deployment, quality, or returns.
Our read
Ramp’s payment data may capture paid adoption sooner than surveys, but the episode does not reconcile the measures. Use its index as a spending signal, not proof of broad or productive use.
Token budgets add a third spending input
Token spend is the operating question behind the adoption numbers. Glyman said AI spending in Ramp’s customer data grew 21 times over the last year, and called tokens a third input beside employees and vendors.
He offered 10 cents, $1, and $20 as examples of what an agentic question might cost. Costs depend on what the session does, including whether agents call other agents.
He said many CFOs lack core capabilities to manage that third spending pillar. That is a governance gap, though he offered no standard unit, policy, or outcome threshold.
Aghi Marietti said on MTS in October 2026 that enterprises were moving across dozens or hundreds of models. He expected token use to keep surging even as inference got cheaper.
Marietti’s point concerns model diversity and volume, not evidence about finance controls. On Latent Space in October 2026, Alex Lieberman estimated his firm’s raw token spend against employee spend.
His estimate was somewhere between one to one and maybe two to one. It is one company’s estimate, not a benchmark for Ramp customers.
- AMAghi Marietti · MTS · Oct 23Enterprise AI is spreading across models as token use surges
Alex Lieberman · Latent Space · Oct 21Tenex's token spend is roughly one to two times employee spend
Our read
Token spend can vary by task and session, making a single AI budget hard to interpret. Track usage, ownership, and outcomes together rather than treating tokens as an automatic substitute for payroll.
Companies are matching model capability to task
Glyman said CFOs were pressing for returns. He then described companies asking whether an Olympic athlete was needed to walk the dog or a Ferrari to pick up groceries.
That distinction matters. CFOs are pressing for returns, while Glyman describes companies evaluating task-to-model fit. The episode gives no measured savings from switching models.
Woodson Martin, OutSystems CEO, offered a separate example on The Cognitive Revolution in October 2026. He said most enterprise workloads do not need frontier models.
Martin described a custom harness and router that cut token costs. Guillermo Rauch, Vercel’s founder and CEO, offered another example on Sourcery in October 2026.
Rauch said open models handled 80% of Vercel’s token traffic. His examples included support tickets, spelling, and data classification, tasks he did not need an Einstein for every time.
Take those as company-specific precedents, not proof of an industry-wide model mix. For buyers, compare cost and output quality on a real task, including errors and escalation.
Cheaper tokens alone do not establish lower total cost if people must repair outputs or workflows stall.
Woodson Martin · The Cognitive Revolution · Oct 7Most enterprise workloads do not need frontier models
Guillermo Rauch · Sourcery · Oct 16Open models now handle most token traffic on Vercel
Our read
Task-level cost and quality comparisons are more useful than a blanket model rule. The examples show what two companies did, not what every workflow can support.
Blackstone’s data shows hiring and AI spend coexisted
The earlier Economic Weather Report said LLM spending across Blackstone’s portfolio companies was up approximately 20 times year over year. That is a separate portfolio-wide spending signal.
Later, Tim Coolican said employment across North American portfolio companies grew 2.7%. No period was specified for that figure in the debrief.
Those figures show AI spending and employment rising in the same portfolio context. But the episode does not link them company by company, or say which roles accounted for hiring.
Aggregate hiring also cannot show whether AI changed staffing decisions. For a board, the practical check is to compare task-level spending, output, and staffing over the same period.
Ramp’s finding supports monitoring adoption and hiring together. It does not settle causation, returns, or labor effects across firms.
Our read
The Blackstone numbers echo the possibility that AI spending and employment can rise together. They do not show a causal link, so keep spend, output, and staffing tied to the same businesses and period.
What it means for you
- Map AI spending. Separate paid adoption from high-intensity use, then record the employee denominator. Ramp’s data shows that spend varies sharply across adopters.
- Test one workflow. Compare model costs with output quality, errors, and human repair time on a real task. Keep a stronger-model fallback for cases that need it.
- Track hiring alongside use. Compare headcount changes across AI-spend groups while accounting for business mix and growth. Ramp’s 12% finding is a reason to investigate, not a planning assumption.
- Assign token ownership. Give teams visibility into agent sessions and their costs. The episode’s examples range from cents to dollars per question.
What to watch
- Whether Ramp publishes more detail on the hiring comparison’s sample, period, and controls.
- Whether AI spending produces measured output gains within specific workflows.
- Whether companies develop task-level controls as token use spreads across models and agents.
Send this to the CFO, people leader, and product owner setting AI budgets.
Watch the minute where Glyman gives the 12% hiring comparison
Written by Real Alpha from the episode's transcript and the other moments it links, then checked against the transcript. Quotes are the speaker's exact words.
Minute by minute
- 4:58Ramp ties growth to savings
Glyman describes Ramp as a platform for smart cards, payments and accounting automation that reduces the need for other tools. He says a typical business can cut expenses by 7.5% per year and argues that tying Ramp’s model to customer savings sets it apart from providers that reward spending more. He says Ramp generates well over $2 billion in annual revenue, is growing more than 100% year over year, and handles about 1% of U.S. corporate payments.
- 7:13Downturns sharpened Ramp’s proposition
Glyman says Ramp launched in 2020, a month before the pandemic, but remote work made expense management more relevant. When interest rates rose in 2022 and competing businesses struggled, Ramp’s focus on reducing waste resonated more, he says. In 2022 and 2023, Ramp raised money at a lower valuation, hired more people and invested in technology, which Glyman says helped it pull ahead.
- 9:00Faster decisions start with fewer meetings
Glyman says he looks at how long it takes employees to buy software, how quickly teams ship code and features, and how much time goes to recurring meetings. Ramp cancels all meetings at the start of the year and adds back the ones teams need, he says.
- 10:19AI adoption varies sharply by company
Using a sample of about 50,000 businesses, Glyman says the top quartile of AI spenders—about 12,500 businesses—roughly doubled revenue over the three years after ChatGPT launched, while the bottom quartile grew about 10%. He says Ramp’s spending data puts AI adoption ahead of survey estimates, with adoption in healthcare, restaurants, food service and manufacturing ranging from 25% to 40%. Glyman says about 55% of businesses in Ramp’s data spend on AI; monthly spend is about $12 per employee at the median and almost $10,000 among the top 1%.
- 14:36AI-intensive firms are hiring more
Glyman says Ramp and Revelio Labs found that firms spending at least $30 per employee per month on AI hired 12% more people than firms that did not adopt AI or adopted it later. He says the finding challenges the expectation that AI adoption necessarily reduces entry-level work.
- 15:42Tokens add a third spending category
Glyman says inference spending is estimated at $160 to $180 billion, compared with $8 to $10 trillion in annual U.S. white-collar labor costs, and that Ramp’s AI spend grew 21 times over the last year. He frames tokens as a third input alongside employees and vendors: companies rent model intelligence through agents, and a question could cost 10 cents, $1 or $20. As CFOs seek returns and choose the right model for each task, he says many lack tools to manage this spending.
- 18:47AI does not erase enterprise software
Glyman calls the economy surprisingly healthy and says more small businesses are starting and succeeding. He argues that enterprise software still helps companies set governance, store information and define business processes. He compares today’s AI models to new employees who need to learn how a company works.
- 20:22Interesting marketing earns attention
Glyman says Ramp’s campaign with Brian Baumgartner doing bill-pay in a glass office for eight hours was its most viral, with an estimated 300 to 400 million people watching. He says the team asks how to be interesting rather than how to distribute a message, and points to Tony Hawk and Bam Margera skateboarding as an example of what draws curiosity.
- 22:08Ramp hires for exceptional strengths
Glyman says conventional job descriptions can filter out people with standout ability in a core task. He cites young engineers who learned to code through Minecraft servers and his brother, who had difficulty with writing but excelled at history and research. He says Ramp looks for overlooked people whose deep focus fits work such as building large-scale systems.
- 25:05Ramp prioritizes lasting growth
Glyman says Ramp hopes to become public one day but is focused on building a durable, cash-generating business. In the debrief, Tim Coolican says Blackstone portfolio companies’ AI spending was 20 times what it was a year earlier; Christine Anderson reports North American portfolio-company employment growth of 2.7%, which Coolican says resembles Ramp’s hiring data. Coolican says Ramp’s focus on saving time and money makes its brand and product coherent.
Published Nov 3, 2026.
Transcript · 23 min read
Inside Blackstone, hosted by Christine Anderson, with Eric Glyman.
From the transcript Inside Blackstone publishes with the episode.
I think so many times in marketing in companies, you have brilliant people. And then the question that gets asked is, how do we get our message out there? And it kills so many ideas, cause you have a message and it's about distribution. And the question the team asks is, how do we be interesting? Who's doing interesting things in the world? That was just Eric Glyman, Co- Founder and Co- CEO of Ramp, a finance automation platform, now worth $ 60 billion. Stay with us to hear what the spending patterns of more than 85, 000 companies tells us about the economy, corporate AI adoption, and employment.
I'm Christine Anderson, Global Head of Corporate Affairs, and this is Inside Blackstone. Here's this week's lineup. First, Winfield Sickles is back with the Economic Weather Report. Next up, the main event, my conversation with Eric. And finally, I invited my colleague, Tim Coolican, to the Debrief. He's an operating partner who works with companies on brand and marketing. We're gonna unpack some of Ramp's very unconventional marketing tactics. Thanks, Christine. Last week was the busiest week of Q3 earnings season with S&P 500 earnings tracking 29% growth and revenues up 14%, both reflecting resilient corporate fundamentals.
That said, equity markets were under pressure last week with earnings from hyperscalers mixed versus expectations, driving the Nasdaq 100 down nearly 10% below its early October peak. Despite this volatility, the key takeaway is that underlying AI demand trends remain strong. Microsoft, Google, and Amazon saw between 40% to 100% cloud revenue growth last quarter, combined with a revenue backlog of nearly $ 2 trillion. The latest macro data tells a similar, positive story. Preliminary Q3 GDP showed the US economy grew an impressive 3. 4%, supported by continued consumer spending and strong business investment, particularly around AI CapEx.
Read the whole transcript
And then against that backdrop, the Fed held rates steady on Wednesday, as expected, highlighting strong productivity growth, CapEx, and economic activity. At Blackstone, what gives us conviction is that we're seeing the same thing in our own portfolio. Revenue growth across our global private equity operating companies grew 10% year- over- year in Q3 with continued margin expansion. Energy was our strongest- performing sector with revenue growth of 21%, followed by industrials at 13%. Both are benefiting from extraordinary AI- related demand. We can also see the scale of that demand directly.
Large language model spending across our portfolio companies is up approximately 20 times year- over- year. And importantly, this isn't just a backward- looking story. Despite all of these noisy headlines, almost 100% of our CEOs, when we surveyed them, expect business conditions to improve or remain stable over the next six to 12 months. So the bottom line is this: we're heading into Q4 with plenty of external volatility, but the signal underneath the noise remains remarkably consistent. Healthy growth, strong corporate fundamentals, and an AI investment cycle that continues to accelerate and is beginning to drive tangible business results.
Looking ahead later this week, in addition to more corporate earnings, we'll get fresh reads on the labor market and business activity. All this will give us more insight into the underlying health of the economy as we head into year end. And now over to Christine's conversation, and I can't wait for this one, with Ramp’s Co- Founder and Co- CEO, Eric Glyman. What is, oh I see your pens! That's what they are. Yeah. Yeah. I'm very nerdy. It’s like, who is this guy? And now for the main event, my guest today, Eric Glyman, Co- Founder and Co- CEO of Ramp. Ramp makes software that helps companies manage expenses.
Sound a little dry? Well, there’s way more here than you might think. Ramp has grown extraordinarily fast, becoming one of the world’s most valuable, private fintech companies. More than 85, 000 businesses use Ramp giving it a fascinating window into how businesses are spending money, including how quickly they’re adopting artificial intelligence. Along the way, Ramp has also made waves with some very unconventional marketing, which of course, I can’t wait to discuss. Eric, welcome to Inside Blackstone. Thanks. So nice to have you here. Thanks so much for having me, Christine. All right, well, you took the subway up to be here.
I really appreciate it. Gorgeous day out. Thank you. Yeah, and public transit is working fast as usual. I'm so glad. I'm so glad. So seven years ago… Yeah. … you found Ramp. Give me the short version of Ramp. If you've ever done an expense report and didn't like it, we might have good news... I have. ... for you. Yeah, well we have good news for you! You can think about it as a better way to run your business. From one place, you can issue smart cards that know your expense policy. They'll do your expenses for you. You can make payments of all kinds, check wires, ACH, and even automate accounting.
And so part of what we're trying to do is have financial products that take care of all of the work in the background. And the effect is, not only do you need a lot fewer tools to run your business, but you spend less. In a typical business is able to cut their expenses by 7. 5% per year.
“What if we designed a business model where the more we help our customers save in terms of money and time, perhaps, the faster we would grow.”
“And one of the things we like to do is, often at the start of the year, we just cancel all the meetings.”
Yeah. A real focus on… Yeah. … prioritization which is a good unlock. Okay, so let's talk about your data. Yeah. So like Blackstone, you sit atop just a tremendous amount of customer data.
“these firms, despite growing at these same rates, were hiring 12% more people than ones who did not or would adopt it later.”
“Many CFOs are lacking kind of the core capabilities to manage this third pillar of spend.”
You see billions of dollars of business spend. What does that tell you about the economy right now? I think the economy is in a surprisingly healthy place. I grew up in Las Vegas, and so, I went to college right as the financial crisis was going on and I think about this a lot. We think that businesses are in a relatively healthy place. And I think that American businesses are more competitive. We also see more new businesses now than we've ever seen. It's so hard to be a small business owner, to have an idea and go out there. Cause you're like, I needa find a lawyer. I needa find someone who knows how to do the books.
I needa find someone else who knows how to do sales. And we're seeing more small businesses start and succeed than we've seen in the past. Okay, talk to me about software. Yeah. Obviously in the news a lot lately, you know, AI, there's all this buzz that AI would be the death of software, but enterprise software seems really here to stay. What are you seeing? I fundamentally believe that, you know, access, at a low cost, to intelligence is gonna create a lot of goodness in the world. It's gonna allow people to pursue things that they want, but so much of what enterprise software is doing is trying to help businesses set governance, storage, define their business processes and how they want things to interact.
People experience this all the time. You hire a precocious college student or high school student to come into your business. They're really smart and they do really dumb things on day one. And it's great, they're gonna get there. I was there too. And they kind of learn the ins and outs, how the business works, why the structure is that way. And we're in that moment right now with these models. Okay, so let's switch to marketing… Yeah. … for a second. Yeah. You have taken a very unconventional approach to marketing, almost like a consumer brand. Yeah. So you had one where you took Kevin from the hit show, The Office… Yeah. … right?
Brian Baumgartner… Yeah. and you put him in sort of like a glass office in the… Yeah. … Flatiron District for eight hours… Yeah. … and had him do sort of, like, bill pay… Yeah. … right. So like where did that come from? And what was the origin of that? It's a… Brian's amazing. Like unbelievable talent… Long day for Brian. It was, you know, I think he said it was one of the most exhausting, but hopefully, I think rewarding days. Like, that was the most viral thing we've ever done. I think, when all is said and done, we had 300, 400 million people watched it. It’s wild. I think so many times in marketing in companies, you have brilliant people.
And then the question that gets asked is, how do we get our message out there? And it kills so many ideas, cause you have a message and it's about distribution. And the question the team asks is, how do we be interesting? Who's doing interesting things in the world? And you start exploring. If you actually do something interesting, if you stand out, people get curious. Wait, what did they do? Tony Hawk and Bam Margera on a ramp? And... Skateboarding. And you start going and so it's, it’s been an amazing way to stand out and I don't know, make people feel something. In something as staid as spend management, how do you start to bring that world to life?
How do you show a wink? Well, I agree the wink is so critical and, you, just hats off to you and your team, I think. I really admire what you've done.
“And so when we're trying to build systems of extraordinary scale, this is who I'm calling first.”
I think there’s a lot of people with great talent that are not given a fair shot to show what they can do. Ramp’s looking at an IPO? Is that in your future? It’s a, you know, one day we aspire to be a public company. You know, today, we’re just tryna build a business that stands the test of time. We generate cash, growing faster, and we intend to be there. Well, it was so nice to have you with us. Thank you so much for being here. Thanks again. I'm joined now by one of our portfolio operating partners, my colleague, Tim Coolican. Before joining Blackstone, Tim spent his career building some of the world's best- known consumer brands, including running L'Oréal Paris in the United States.
Today, he works with our portfolio companies on brand, marketing, and growth initiatives. Tim, welcome to Inside Blackstone. Thanks so much for having me. All right, well, what a captivating guy, Eric. Fascinating on so many fronts, but I wanted to talk to you about all things brand, but maybe also talk a little bit about their data. And how that syncs with what we're seeing across our portfolio. So first, AI adoption. What’d you think of what he said? We're seeing exactly the same thing, you know, the pace of adoption is moving at light speed. And as Jon shared in the Q3 earnings, our companies are spending 20 times what they were a year ago.
The adoption is moving very fast. And it's all about turning that adoption into productivity and real business results. And we're really seeing encouraging signs. Well, it's so interesting what their data is showing about AI adoption and what that means for employment numbers. Yeah, I thought that was really interesting cause it challenges the conventional narrative that AI adoption equals job loss. And in fact, our employment numbers grew by 2. 7% across North American PortCos. So our data is showing something pretty similar. 100%. All right, now let's get to brand. You know, so interesting what Ramp is doing. When you think about it, B2B company sort of acting more like a B2C company with all of these marketing stunts and tactics that they've tried. What's your take?
I think it's fascinating because he really takes the perspective of the consumer and of how the brand is going to connect to the consumer in everything he does. So their why is so clear, who doesn't want to save time and money. It's an eternal need. And the way that he thinks about it, through the product, it's really delivering and delivering in a really engaging way. When he talked a little bit about, you know, how do we be interesting, is the first question he asked, not sort of, how do get our message out, which is a little bit different. I love that framing. We're in a new era of how people discover and how they make decisions.
And there are a lot of ways to reach people. It's really hard to get their attention and to keep it. So I think the fact that Eric has taken that perspective of, how do we get people to care, it's unlocking so much creativity and that creativity is driving interest and engagement and ultimately, trust and loyalty. Okay so when you look at Ramp’s tactics, what do you think are the key lessons from them? I think the greatest brands are inside out. And with Ramp, their mission and their why is so clear, to save people time and money, and who doesn't want that. And their product does exactly that in a really unique way.
So, who they are on the inside is what you see on the outside. It’s kind of like Inside Blackstone. Yeah, absolutely. I love that setup. All right, well, thank you so much for joining us. I hope you’ll come back. Thank you so much. And thank you for joining us on this week's episode of Inside Blackstone. Remember to follow us and subscribe wherever you get your podcasts.
