How Ramp CEO Eric Glyman Is Using AI to Build “Self-Driving Money”

TL;DR

Ramp CEO Eric Glyman describes self-driving money as zero-touch financial automation that handles work invisibly instead of making users interact with chatbots. Ramp provides one system for issuing cards, managing approvals, making payments, and automating book closing, while its average customer saves about 5% per year on expenses. Read on to see how Ramp applies this vision and what Glyman learned from building an AI agent in 2015.

Transcript

look I think the most important thing um Beyond is empathy if you're trying to make great products you need to have great taste um you need people who are making to so deeply understand um really the experience of a building the pain that people are going through um you know what they're actually you know the customer at the end is actually doing o... Read More

Key Insights

• Zero-touch automation that works invisibly in the background is more valuable than chatbots for many finance tasks, because few people actually want to chat with their bank account or expense report, and the interaction adds friction rather than removing it.
• The average company using Ramp saves about 5% per year on its expenses, which over five years in business has added up to billions of dollars of savings and the equivalent of thousands of years of saved labor.
• Ramp is a command-and-control system for company finances where, from one place, businesses can issue cards, manage approvals, make payments of all kinds, and even automate closing their books.
• Great products require empathy and deep understanding of the customer's pain, in some cases better than the customer understands it themselves, which is why Ramp engineers join customer calls and are accountable for metrics when shipping.
• Agentic AI is not new: in 2015 at Paribus, Glyman built an email-integrated agent that scanned inboxes for receipts, tracked prices, and auto-generated emails to stores to claim price-drop refunds, used by millions.
• Ramp applies large language models to transaction data and the general ledger to autocomplete expenses faster and, for most customers, more accurately, effectively letting a company's books do themselves.
• Zero-touch expenses replace two apps with one: instead of using American Express plus Concur, tapping a Ramp card pulls the receipt from the merchant or email so the expense report is done for you.
• Ramp considers itself more a productivity company than a money company, so almost anything that raises the productivity of capital or saves finance teams time or money is in scope for its self-driving money vision.

Questions & Answers

Q: What is self-driving money, according to Ramp CEO Eric Glyman?

Self-driving money is Ramp’s vision for software that performs financial work automatically on a company’s behalf. It targets tedious tasks such as auditing expense reports, classifying vendors, and maintaining records, with zero-touch automation operating invisibly in the background.

Q: What does Ramp do for companies?

Ramp is a command-and-control system for company finances designed to make businesses more profitable and help them operate more smoothly. From one place, companies can issue cards, manage approvals, make different kinds of payments, automate business processes, and even automate closing their books.

Q: Why does Eric Glyman prefer zero-touch automation to finance chatbots?

Glyman says few people want to chat with a bank account or expense report. He believes chat interfaces can add friction, while zero-touch AI can complete finance tasks invisibly and automatically, making self-driving cars a better analogy for Ramp’s approach.

Q: How much does the average company save by using Ramp?

The average company using Ramp saves about 5% per year on its expenses. Across Ramp’s first five years in business, customer savings totaled billions of dollars alongside the equivalent of thousands of years of saved labor.

Q: How does Ramp use large language models to automate expenses?

Ramp applies large language models to transaction data and a company’s general ledger to autocomplete expenses. For most customers, this completes the work faster and more accurately, helping their books effectively do themselves.

Q: How do Ramp’s zero-touch expenses work?

Instead of combining a payment card with a separate expense-reporting app, a user taps a Ramp card and the system retrieves the receipt from the merchant or the user’s email. The expense report is then completed for the user rather than manually prepared by them.

Q: What AI agent did Eric Glyman build at Paribus in 2015?

Paribus built an email-integrated agent that scanned inboxes for purchase receipts, tracked prices, and detected eligible price drops. It generated refund-request emails to retailers, allowing users to recover differences such as $100 when a $1,000 television dropped to $900; millions of people used the service.

Q: Why does Eric Glyman say customer empathy matters when building AI products?

Glyman says product teams must understand customers’ experiences and pain deeply, sometimes better than customers understand them themselves, before they can design effective automation. At Ramp, engineers join customer calls, remain accountable for product metrics, and regularly speak with customers.

Summary & Key Takeaways

• Ramp is dedicated to making companies more profitable and operate more smoothly through a command-and-control system for company finances. From one place, businesses issue cards, manage approvals, make payments, and automate closing their books. The average customer saves about 5% per year, totaling billions in savings across 25,000 businesses including Shopify and Virgin Voyages.

• Eric Glyman built an early AI agent in 2015 at Paribus that lived in your inbox, scanned for receipts, tracked prices, and auto-generated emails to stores claiming price-drop refunds. Millions used it, proving agentic AI existed a decade ago, though it was a narrow use case compared to today's more generalized reasoning capabilities.

• Ramp's vision is self-driving money: taking tedious, monotonous finance tasks like auditing expense reports by hand or tagging vendors and automating them. Glyman rejects the post-ChatGPT rush to add chatbots everywhere, favoring zero-touch AI that works invisibly. He views Ramp as a productivity company, so almost anything saving time or money is in scope.