Keith Rabois and Eric Glyman on Ramp, Execution and Building Momentum

Harry Stebbings talks with Ramp co-founder and CEO Eric Glyman and investor Keith Rabois. Rabois says a Founders Fund colleague met Glyman’s co-founder Kareem through gaming and introduced them. The founders flew to San Francisco quickly, and their 2019 pitch matched the investors’ view of the opportunity in business finance.

Rabois says Khosla Ventures later became interested in Ramp partly because of its potential to apply AI to finance and his own involvement. He led an earlier Founders Fund investment and was involved with Khosla’s later investment, so his assessment of Ramp is also an investor’s commercial perspective.

Glyman describes Ramp not simply as a corporate card but as a productivity and workflow business. He says it connects spending and other data to identify unused software seats, suggest accounting categories and automate expense processes. He presents saving customers money and time as the core value proposition; those performance claims are company statements, not independently verified results.

Rabois credits Glyman with an early focus on talent and strong marketing instincts. He recalls a first board meeting spent substantially on prospective hires and says the investors preempted the Series A before the company had meaningfully launched. The two guests praise each other’s work; Rabois declines to identify a major weakness when pressed.

Glyman credits Rabois with teaching him to map a business as inputs and outputs. The exercise forces leaders to state what they are trying to change, identify the levers that drive it and focus the team. He also points founders to Rabois’s operating framework and emphasizes talent, clarity and consistent execution.

In Ramp’s early card model, Glyman says, the team mapped purchase volume, interchange and funding costs. They concluded that purchase volume influenced all three: it increased interchange and could improve funding terms and fraud prediction. Their question became how to build a product that companies wanted to use deeply, rather than simply acquire a first card customer.

Glyman says a small company might prioritize card points and lounge access, while a larger one increasingly needs predictable costs and easier expense and accounting workflows. Ramp targeted being a company’s last card rather than its first. He says customer spending can rise much faster than headcount, making this an attractive business model; the examples are his observations, not a universal growth rate.

Ramp counts the days since founding. Glyman says the practice is not a call to rush blindly but a way to reflect on time: at the first two board meetings, the count moved from 133 to 199, prompting the founders to ask what they had accomplished in those 66 days. A calendar audit helps them decline lower-impact work and spend scarce time where it produces leverage.

Asked about missed opportunities, Glyman says Ramp has generally preferred to launch, test and learn quickly. He notes that it historically did not hold customer funds and might have benefited from doing so during the prior year’s bank failures. But he values the consistency of its mission to save customers time and money rather than changing direction for every opportunity.

Rabois sees organizational design as a central scaling challenge: decision-making and responsibilities that work at one size can fail when headcount multiplies. Glyman adds the challenge of building a nationally recognized brand and expanding beyond cards. His estimate that Ramp represented about 1% of how US businesses spend on cards is a company estimate with a specific market scope.

Stebbings asks whether a company needs repeated executive turnover as it grows. Rabois says some changes can be appropriate, like bringing in a relief pitcher with different strengths, but rejects a fixed requirement for five turnovers. Glyman prefers to stretch strong existing employees into larger roles when possible, while acknowledging the need for outside expertise.

Rabois describes an internal-promotion culture at PayPal and suggests a mix of internal growth and selective outside hires, rather than a universal formula. To assess leaders, he borrows Brian Chesky’s six-month-ahead test: a leader should not merely deliver today’s results but anticipate issues early enough to adjust the levers that take time to change.

Jack Dorsey’s editing metaphor gives Rabois another test. A CEO should generally edit a leader’s work rather than write it from scratch. Simplifying can be useful, but repeatedly making extensive corrections or asking basic clarifying questions in the same area may signal a leadership problem.

Whether to intervene closely or delegate depends, Rabois says, on task-relevant maturity and the consequences of failure. A person facing a new problem with high stakes may need more frequent check-ins than someone experienced in a familiar task. The CEO remains accountable for the result even when responsibility is delegated.

Glyman says founders should credit the team for successes and take responsibility for problems. Ramp begins board materials with what is going well, what is going less well and why. He wants people to surface misses without hiding them or punishing colleagues, so leaders can compare their beliefs with reality and correct course.

Asked whether great CEOs are defined by a few big decisions or many smaller ones, Glyman leans toward repeated decisions and placing hours where they have the most leverage. Rabois adds that startups must overcome inertia and build momentum, then recognize promising opportunities and reinforce them.

Rabois says founders can create an initial spark with activities that do not scale, but cannot rely on shortcuts forever: a real engine and repeatable process must follow. Publicity or friends and family sometimes help start momentum, though neither is a reliable formula. He stresses that the path is different for each startup.

Rabois says early-stage investors will miss more often than they hit and acknowledges screening out people he should have met. As an executive, he argues, the tolerance for poor hiring decisions is different: a bad hire can undermine a team and cost significant momentum. He does not advocate an impossible goal of zero hiring mistakes.

Glyman advises founders to start with curiosity about a durable customer problem rather than begin with a new technology and search for an application. Ramp’s founding premise, he says, was that companies want to spend less, become more profitable and save time; the technology followed that mission.

He traces this concern to Paribus, his earlier company acquired by Capital One, which helped consumers seek refunds after price drops. At Ramp, customer feedback broadened the focus from saving money to saving time on expenses and closing books. Glyman reports revenue well into nine figures and says Ramp had less than 1% of its largest market; the period and market definition are not specified in that statement.

Moving from a product to a platform creates a sequencing challenge, Rabois says. A clear product promise can become harder to preserve as the offering broadens. He argues that integrated data can unlock more valuable workflows, but that the team still needs to test the value and prioritize among opportunities.

Asked whether finance is a monopoly market, Rabois argues that future CFOs will want Ramp because he expects it to make them more effective. He cites an estimate that finance teams spend 17% of their time on strategic analysis, while Glyman cites an approximate 5% of finance roles as strategic. Neither supplies supporting evidence in the interview, and both have a direct interest in Ramp’s success.

Rabois imagines a future Ramp that helps customers allocate resources, apply policies and use integrated information for strategy. These are ambitions for the product, not evidence that it currently performs fiduciary or compliance functions autonomously. Glyman says the company still has a long way to go.

In a company postmortem exercise, Rabois identifies hiring mistakes and getting financial risk wrong as potential threats. Glyman adds a loss of focus: expanding to many connected workflows must not come at the expense of excellence in individual products. Both frame these as risks to manage, rather than reports that they have occurred.

Glyman says one-on-one meetings occupy much of his calendar and he plans to shorten or skip some. Rabois reviews his calendar weekly and says there is no universal number of direct reports: the right span depends partly on the leader’s familiarity with each function and the amount of preparation needed.

The guests discuss the round’s roughly $7.6 billion post-money valuation. Rabois says he does not model a prescribed three-, four- or five-fold return; instead, he asks whether the vision could make the company useful to businesses broadly and therefore potentially much more valuable. This is his investment thesis, not a promised return.

Looking ten years ahead, Glyman wants Ramp to make customers more efficient and profitable. He says the company currently saves the average customer 5%, and recalls about 2% on card expenses at launch; the interview does not establish the methodology or independently verify those figures. His hopes of larger savings and much wider adoption are future goals, not results.