The $60M seed round is the tell — enterprise IT just realized they have no idea which AI agents are accessing their crown jewels.
The Summary
- Oak Inc. raised $60M in seed funding from Accel, Greylock Partners, and CRV to solve enterprise identity management as AI agents multiply across corporate networks
- Companies can't track who (or what) accesses applications anymore, creating a massive security gap in the agent economy
- The seed size signals how urgent the problem is: identity infrastructure built for humans breaks when non-human agents outnumber employees
The Signal
Oak launched with the kind of seed round that usually comes after you prove the market exists. They didn't need to. Every CISO already knows the problem: their access logs look like chaos because AI agents don't fit the identity model built for humans with email addresses and two-factor auth.
The timing matters. As companies deploy agents for customer service, data analysis, code generation, and procurement, those agents need access to internal systems. Your sales agent needs the CRM. Your finance agent needs the ERP. Your coding agent needs GitHub and production databases. Traditional identity and access management systems were built assuming every entity with credentials is a person who logs in once a day, not an agent that makes 10,000 API calls an hour.
"Identity infrastructure built for humans breaks when non-human agents outnumber employees."
The enterprise access gap Oak is targeting comes from three colliding forces:
- Agent proliferation happening faster than IT can adapt policies
- Legacy IAM systems that can't distinguish between employee, contractor, and autonomous agent activity patterns
- Compliance requirements that still assume humans are the only entities touching sensitive data
Oak is betting that Web4 infrastructure needs Web4 identity. The companies writing $60M seed checks agree. Accel and Greylock don't move this fast on problems that might emerge in three years. They move this fast on problems that are already bleeding enterprises money through security incidents and compliance failures.
The identity layer is where the agent economy either scales or stalls. If you can't audit which agents accessed what data and why, you can't deploy them in regulated industries. If you can't revoke agent permissions as granularly as you revoke human access, you can't contain breaches. Oak is building the connective tissue between the old world of human identity and the new world where most entities on your network aren't human.
The Implication
Watch for Oak to become infrastructure every multi-agent deployment depends on. If they execute, they become the identity backbone for Web4 the same way Okta became it for Web2. The strategic question for enterprises: are you waiting for your IAM vendor to bolt on agent features, or moving to purpose-built solutions now.
For anyone building agent platforms or selling AI into the enterprise, Oak's funding is a signal about what features you need to ship. Logging, auditing, and granular permission management for non-human entities just became table stakes.