AI Startups Need Enterprise Controls
Michael Grinich, CEO of WorkOS, on AI startups getting enterprise-ready at launch
The product did not fail because users did not care, it failed because the buyer changed once Nylas tried to sell beyond enthusiasts. Email software touches a company’s most sensitive data, so adoption quickly moved from an employee choice to a review by IT, security, and procurement. That meant the roadmap had to include SSO, auditability, directory controls, and contract workflows, not just a better inbox experience.
-
This is the core split between user love and enterprise revenue. A team can start using a product bottom up, but once seat count or sensitivity rises, the deal becomes an invoiced purchase with security review, procurement negotiation, and legal terms that cannot be self served.
-
Email hit that wall earlier than many SaaS categories because it is already the system of record for customer conversations, internal approvals, and identity. That is why adjacent winners like Front built collaboration into the inbox, while WorkOS was built to supply the enterprise controls that products need to survive those reviews.
-
The strategic lesson became WorkOS. Instead of asking every startup to spend a year building SSO, directory sync, audit logs, and other enterprise blockers from scratch, WorkOS packages those features as APIs so a small product team can become enterprise sellable much earlier.
This pattern is moving faster in AI than it did in the earlier PLG era. Startups now reach enterprise demand within months, not years, so the companies that standardize security, identity, and procurement readiness are becoming part of the launch stack for any product that wants to turn early usage into durable commercial accounts.