How AI Is Repricing Every Software Company

She says customers quickly recognize when an app is just reselling model access, comparing the implied per-token price to buying model tokens directly and asking whether the app is really “4X as good as Claude.” In her view, this forces the application into an unwanted head-to-head comparison with the underlying models, which is hard to win if there’s any markup.

She describes Vanta’s product experience shifting from monitoring/alerting toward an “agent” that proactively proposes actions and, with user permission, executes remediation steps (e.g., offboarding access when someone leaves a company). The thrust is moving from “here’s a problem” to “I fixed it—approve?” to keep users in a productive flow loop.

She argues the most durable advantage has been relentless shipping plus a large integration footprint—connecting to ~400 tools—and using that data foundation (what she calls a “trust graph,” effectively a security data lake) to power automation and new capabilities. She frames this as a compounding platform that lets Vanta build faster than competitors.

She describes “cognitive dissonance” in an environment where high-growth, high-margin SaaS businesses are valued at far lower multiples than before, while lower-margin, higher-growth profiles can be rewarded more. Beyond valuation, she emphasizes second-order effects: fundraising narratives shifting toward “AI maximalism” and heightened retention/hiring pressure from the talent market.

She says many teams default to familiar patterns (like onboarding wizards) because they know they’ll work and fear failing on critical projects. In the example, Vanta replaced wizard-style onboarding with an LLM-style “canvas” where users drag in messy files and screenshots, and the system makes sense of them through questions—after leadership explicitly “burned the boats” and accepted risk.