InsightfulDiscussion

How To Build The Future: Eric Glyman, Co-CEO of Ramp

Y Combinator

Eric Glyman, co-CEO of Ramp, discusses how the company evolved from Parabus to build a $60 billion financial operations platform, and how Ramp is leveraging AI agents and internal tools like Inspect, Research, and Glass to transform organizational productivity and reshape the future of work.

Summary

Eric Glyman shares the journey of building two successful companies and the lessons learned along the way. He and co-founder Kareem first created Parabus, a savings app that automatically found price drops and requested refunds—essentially an early AI agent. After being rejected from Y Combinator initially, they applied again and were accepted in Summer 2015, eventually growing to nearly a million daily users before selling to Capital One. At Capital One, Glyman observed how organizational incentives often misaligned with customer value, with financial institutions designing products that subtly harmed customers to benefit the company.

This insight led to founding Ramp with a core philosophy: design financial products to genuinely help companies spend less, competing on value rather than price. Ramp started as a corporate card and expense management platform, and has grown to serve tens of thousands of businesses with a $60 billion valuation. The first chapter of Ramp focused on creating a unified financial operations layer, consolidating tools for credit cards, expenses, bills, procurement, and travel. Median Ramp customers now save 7.5% annually on expenses.

The second chapter, emerging now, involves leveraging AI to drive productivity and growth. Glyman describes how AI capabilities have advanced dramatically—from GPT-3.5 struggling with basic arithmetic to current models solving century-old mathematical problems. However, he notes a critical bottleneck: the top 1% of AI token spenders consume 80% of spending, while the remaining 99% struggle with adoption due to integration challenges, learning curves, and organizational resistance.

Ramp has built internal AI tools to solve this problem: Inspect (a coding agent handling 85% of pull requests), Research (an internal AI analyst), and Glass (a company-wide knowledge and workflow system connecting all data sources with proper access controls). Glass saw near-universal adoption at Ramp (1,925 of 1,930 employees used it in 24 hours) and enabled radical productivity gains—engineers became 100x more productive, designers 2x more productive, with engineers now creating 7,000 pull requests per quarter, mostly by non-engineers using Inspect.

Glyman discusses how this created organizational friction when different functions adopted AI at different rates, requiring a hiring freeze to allow time for everyone to reach "aha moments" with AI rather than treating it as a 5% time allocation. He emphasizes that organizations must be reimagined for the AI era, noting that large companies can now move with startup-like speed.

On the broader AI future, Glyman believes in the "hyper-specialization" scenario where thousands of specialized billion-dollar companies flourish, rather than a winner-take-all outcome. He notes that Jevons Paradox—where cost reductions drive increased consumption—recently broke for AI usage, suggesting adoption has plateaued despite price drops, likely due to organizational friction rather than technical limitations.

Ramp is beginning to release these AI harness tools to customers, including Glass access, with Model Context Protocol (MCP) usage growing 10% weekly. The company achieved 25% accelerated growth rates after implementing internal AI tools, while reducing expense waste from 5% to 7.5% annually for customers.

Glyman concludes by emphasizing the importance of asking for help as a founder and leader—a lesson from his YC experience that enabled much of his success. He advocates for founders to openly discuss problems and seek advice rather than projecting false confidence.

Key Insights

  • Financial institutions design products that subtly harm customers to benefit the company due to misaligned incentives, where different departments optimize different metrics without coordinating toward customer value
  • The top 1% of AI token spenders consume 80% of spending, while the remaining 99% struggle with adoption due to integration challenges and organizational resistance rather than technological limitations
  • Inspect coding agent handles 85% of Ramp's pull requests with number of requests per engineer growing 40% monthly, with 7,000 pull requests per quarter created by non-engineers, demonstrating radical democratization of technical work
  • Jevons Paradox—where lower costs drive increased consumption—broke for AI over the last two months, suggesting adoption has plateaued despite price drops, indicating organizational friction is the limiting factor not cost
  • Organizations adopting AI at uneven rates across functions creates imbalance where highly productive engineers outrun designers and product teams, requiring hiring freezes and time for all functions to reach adoption parity

Topics

AI adoption and organizational transformationBuilding financial technology platformsAI productivity tools and agents (Inspect, Glass, Research)Parabus to Ramp journey and company pivotsAligning organizational incentives with customer valueAI capability advancement and integration challengesHiring and talent identification strategiesFuture business models and competitive dynamicsInternal AI tool implementation and adoptionBreaking Jevons Paradox in AI consumption

Transcript

[0:00] Today on How to Build the Future, I'm sitting down with Eric Glimman, co-founder and CEO of RAMP. RAMP started seven years ago as a corporate card with one promise. Save companies time and money. Today, it's a $60 billion platform that runs the financial operations of tens of [music] thousands of businesses. And increasingly, it does all of that work using agents. >> We're sort of at a pace [music] that I think most people would look at and go, "This is kind of remarkable." This is the story of [music] how Ramp is reshaping the way companies are built [0:31] from the ground up. [music] Thanks for being here, Eric. >> Honestly, Gary, it's great to be…

Full transcript available for MurmurCast members

Sign Up to Access

More from Y Combinator

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.