AI summary by TwinMind

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

Sequoia Capital

38m 48s

Product Philosophy and Empathy

  • Most important thing beyond empathy: need great taste for product development
  • Must deeply understand customer experience and pain points
  • Need to understand customer needs better than customers themselves in some cases
  • Only then can build well-designed products that automate tasks efficiently
  • Engineers at Ramp participate in customer calls during product releases
  • Engineers accountable for metrics and performance outcomes
  • Everyone at Ramp talks to customers with recent frequency
  • Core part of great product cultures

Ramp Company Overview

  • Mission: make companies more profitable and operate smoothly
  • Built command and control system for company finances
  • One place to: issue cards, manage approvals, make payments, automate book closing
  • Average company saves ~5% per year on expenses using Ramp
  • Over 5 years: billions in savings, thousands of years of labor saved
  • Customers: large public companies (Shopify, Virgin Voyages, Boys and Girls Club) + 25,000 other businesses
  • Focus: make people more productive with time and money

Early AI Experience (2015)

  • Built AI agent at Paribus in 2015 (pre-AI hype)
  • Integrated with email (Gmail, Yahoo) to scan receipts
  • Tracked prices of purchased items
  • Generated emails to stores for price adjustment refunds when prices dropped
  • Users would wake up to $100+ refunds automatically
  • Narrow use case but millions used it
  • Demonstrates agentic AI existed long before current wave

Current AI Vision at Ramp

  • Dedicated to helping spend less money and time
  • AI enables new tools for this mission
  • Focus on pain points: processes causing excess time/money spending
  • Finance teams have tedious monotony: auditing expense reports by hand, downloading spreadsheets, rerunning analyses
  • Vision: automate tedious/monotonous tasks through better design and integrated tools
  • Zero-touch expenses: tap Ramp card → receipt pulled from merchant/email → expense report done automatically
  • Apply LLMs to transaction data and general ledger for autocomplete expenses
  • Books effectively do themselves
  • Over time: point to ways for business to operate more efficiently

AI Implementation Categories

Zero Touch AI

  • No UX interaction required after setup
  • Examples: expense report automation, accounting completion
  • Predicting user inputs with high confidence
  • Expanding surface area of automated tasks

Agentic AI

  • User prompts desired outcome
  • AI shows how to accomplish task and does it
  • Example: "Issue card with $50 budget only for Starbucks"
  • Tour Guide feature: LLM reads screen, guides through steps
  • Built on GPT-4o multimodal capabilities
  • Interface like hiring analyst: "I want this done, go figure it out"

User Experience Philosophy

  • People trained to learn apps vs apps adapting to people
  • Should focus on running business, not managing app intricacies
  • Great design reduces steps, makes things easier/intuitive
  • Balance powerful tools with simplicity
  • Moving toward prompting outcomes rather than navigating interfaces
  • Analogy to self-driving cars: early monitoring → full automation as trust builds

Culture and Customer Focus

  • Not AI-driven finance tool marketed as AI
  • Lead with benefits and outcomes, not technology
  • Impossible to use Ramp without AI (embedded throughout)
  • Focus on problem first, then solution
  • Prevent abstracting makers from problems
  • Everyone talks to customers regularly (unlike many scaling companies)
  • Engineers on customer calls during releases
  • Accountable for metrics and performance

AI as Step Function Change

  • Biggest productivity shift in lifetime
  • Human-level reasoning available through APIs
  • Profound ramifications for software businesses
  • Changed traditional moats (features, integrations, lock-in)
  • AI automation in multiple areas: outreach, SDRs, customer service, copywriting
  • Can listen to 100,000 sales calls simultaneously
  • Most people experimenting with ChatGPT but not incorporating into daily work

Future of Finance Leaders

  • Automate low-value parts of job
  • Current reality: first 6-15 days of month on tagging transactions, downloading spreadsheets
  • Only last 5 days on real strategic work
  • In 5 years: work streams more strategic, insightful, value-creation focused
  • Historical precedent: bookkeepers declined ~1M, financial analysts/strategists grew ~1M
  • People moved to higher abstraction level, more valuable work

Founder Advice

  • Focus on timeless truths that won't change in 10-100 years
  • Start with customer problems that are timeless
  • Real customers, real businesses, actual problems
  • Then discover technological shifts to solve in new/disruptive ways
  • Be timeless rather than timely
  • Great companies built over decades
  • Hopes Ramp is last company he works on

Rapid Fire Responses

Favorite AI App

  • Cognition Labs' Devon (AI engineer)
  • Agentic use case with core innovations
  • Has tools engineers have: notebook, planner, browser, shell access
  • Can watch what Devon does, mistakes it makes
  • Taught new way to think about design

Industries to Change Most

  • Healthcare: continuous measurement vs annual checkups, diagnosis improvement, return humanity to care
  • Design and creation: no longer about understanding tools, but making interesting/intuitive things

Financial Services Observations

  • Traditional players (AmEx, Chase, Citi) founded by people in top hats
  • Never really changed through technology evolution
  • Banks had monopoly on moving/storing money
  • Now non-banks can introduce great technology
  • Apple as fintech innovator: connecting money movement to identification, reducing fraud/friction
  • Disruptors likely from outside traditional finance

Person Most Admired in AI

  • Satya Nadella (Microsoft CEO)
  • Partnership with OpenAI, moving from research to production
  • Thinking in decades, not just weeks/years
  • Sincerity and kindness in operations
  • Mentorship to Ramp team on agentic AI design

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome