eric glyman

Ramp processes over 100,000 expenses daily via LLM agents that interpret plain-English policies, access full transaction metadata, and provide audit trails — achieving 99%+ accuracy vs error-prone human reviewers, turning compliance drudgery into automated separation-of-duties.

As token costs plummet, high-leverage knowledge work (expenses, accounting, procurement) can be delivered for pennies but worth dollars to customers — creating a business model where capturing a slice of massive customer ROI beats fighting over basis points of interchange.

Cursor and Claude Code enable every role (design, PM, support) to ship production code, shrinking problem-to-fix half-life to minutes and potentially rewriting codebases annually — but real companies face HR/tech-debt constraints that slow gains versus greenfield startups.

Historical, hard-to-collect datasets (legal records, WHOIS history, ADS-B feeds) create defensible moats because 10,000 AI geniuses cannot recreate time-travel data collection — making 'boring' data businesses the new oil for vertical AI applications.

Real-time payment data reveals far faster AI tool adoption (ChatGPT, Anthropic, Cursor, Cognition) across mainstream businesses than official surveys capture, suggesting productivity gains and revenue acceleration are already underway but underreported.

Ramp Treasury's rapid growth (billions in deposits) proves businesses will move cash to platforms that auto-sweep to highest yields and integrate with payables — forcing 0.07% checking yields upward as stablecoins and neobanks enable frictionless capital allocation.

AI can clone UI skin-deep but not the nine million edge cases (power outages, regulatory nuances, integration quirks) that constitute a platform's 'dark matter moat' — the best enterprise SaaS (Workday, NetSuite) have hostages, not customers, and operate in a Goldilocks zone of pricing and complexity.