How To Build The Future: Eric Glyman, Co-CEO of Ramp
Y Combinator
Origins at Parabus 0:00
Eric Glyman says Parabus began with a simple idea: if something you bought online dropped in price, a computer could find the difference and ask for a refund for you. The product linked to Gmail or Yahoo, scanned receipts, checked prices, and emailed refund requests. They launched, applied to YC twice, and quickly found users. Within a year, they had nearly a million daily users and were bought by Capital One, where they saw how a large financial company worked and where its incentives often missed the customer.
From refunds to Ramp 8:31
At Capital One, they learned that credit cards and other financial products were often built to grow profit first. Glyman says the real opening was to compete on value, not price, by helping customers keep more money in their accounts. With Ramp, that meant products like cards that also handle expense reports and bill payments. The deeper shift was to move from only spotting waste to taking action on it, so the software could stop money from leaving instead of merely reporting it.
Future first 14:01
Ramp chose startups first because they become larger customers later. At a $60 billion valuation, Glyman calls this “the end of the beginning.” The first chapter was about a single finance layer that replaces scattered tools for cards, expenses, bills, travel, and procurement. The result is simpler workflows and a median 7.5% yearly savings on expenses.
AI changes work 16:31
The next chapter is about helping money and work go further, with accounts receivable and new AI tools like Glass, Inspect, and Research. Glyman says AI moved from a toy that could not do arithmetic to a force reshaping organizations. Inside Ramp, a few people using these tools became far more productive, so the company paused hiring and pushed everyone to learn them.
Measuring outcomes 28:31
Ramp is moving past simple spend tracking. It wants to measure what the spend produced, in one place across all the tools a company uses. The example is plain: if you spend $5 on a landing page and that page converts 3% better, you can tie that to $5,000 more a month. The point is to connect costs back to results, not just count who spent what.
Glass and company memory 29:31
Glass is described as the company brain. It connects to core systems like data warehouses, analytics suites, and Zendesk, then gives people access based on permission. It also captures useful ways of working as shared skills, so one team’s better method can become the default for others. Leaders can see standard procedures, spot differences, and keep the knowledge even when a strong employee leaves.
AI and faster building 43:30
The cost of intelligence may keep dropping until it is almost free, which could turn impossible ideas into new businesses. Glass shows what that looks like at Ramp. Non-engineers created 7,000 pull requests in the second quarter, up from almost none two years ago. Work that once took months can now take minutes. Support staff can fix a customer problem the same day. That raises expectations, speeds releases, and lets Ramp use AI as a productivity tool, with customers now being onboarded onto Glass for their own companies.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.

