Platform controls for the agent era
Vercel has launched Vercel for Enterprise Apps and Agents, a set of platform-level controls aimed at letting entire organizations ship internal applications and AI agents behind their own security boundaries. The offering grew out of Vercel's own experience: employees at the company have built hundreds of internal agents and apps on top of the company's Agent Stack, which surfaced practical questions about governance once those tools were in active use.
The core issues were ownership and access: who may use each agent, how to keep internal tools internal, which data and systems an agent can reach, and what models are being used and at what cost. Vercel for Enterprise Apps and Agents is designed so that answers to those questions become defaults baked into the platform, rather than configuration tasks left to individual builders.
Passport locks internal deployments to your IdP
Early internal agents start as tools for employees, not end users, and that pattern holds across many Vercel customers. The problem was that on a platform built for fast public deployment, "internal" was previously a per-project setting. A single misconfigured deployment could expose company systems and data.
Vercel Passport, now in public beta, changes that default. Every app and agent deployment is private from creation, with access authenticated against your employee identity via Okta, Microsoft Entra, Auth0, or any OpenID Connect-compatible provider. Administrators configure the identity provider connection once, and that policy applies across all deployments. Each entry into a protected app is auditable, and access policy is set centrally, not left to individual developers to get right.
Connect gives agents short-lived access to your stack
Passport governs who reaches an agent. But the agents themselves need access to data and systems to be useful, and that access historically meant long-lived credentials in environment variables, scoped broadly enough to cover anything the agent might eventually do.
Vercel Connect consolidates OAuth, OIDC, and secret injection into a single product that replaces static keys. An agent requests short-lived credentials as it works; tokens are granted per task rather than once and forever, and they expire when the task completes. Connect currently provides access to Slack, GitHub, Snowflake, Salesforce, and Linear, as well as other systems reachable via OAuth or API.
Connect and Passport are both in beta.
Managed users keeps identity in one place
When every employee is a potential builder, account sprawl becomes an operational problem: seats provisioned by no one, stale access after role changes, and no unified record of actions across the platform.
Enterprise Managed Users, in private beta, runs on SAML SSO and Directory Sync to give administrators full lifecycle control over every Vercel builder. Accounts are provisioned automatically from your existing directory, and off-boarding removes access when the directory says so. Group-based access, deployment protection, and MFA enforcement apply org-wide, and all actions land in a single audit trail. The identity provider you already run, whether Okta or another SAML/OIDC provider, now governs Vercel and v0.
Data apps without the ticket queue
v0, Vercel's AI app builder, now connects directly to Snowflake. The intent is to let anyone in an organization safely build data applications on top of the warehouse without an engineering ticket. Access to both v0 and Snowflake is controlled through your identity provider, so data remains internal. You decide who gets a seat, and the apps they build can deploy straight into your Snowflake account.
Platform components | Security implementation |
|---|---|
Vercel Passport | Puts every internal app and agent behind your identity provider by default |
Vercel Connect | Gives agents short-lived, scoped credentials for the systems they use, like Slack, GitHub, Snowflake, Salesforce, and Linear |
Enterprise Managed Users | Full lifecycle control over every Vercel and v0 user through your existing directory |
Bring your own cloud on AWS | Runs apps and agents inside your own AWS account (currently in Private Beta) |
Running inside your own AWS account
For large enterprises, the boundary question extends beyond application privacy to the underlying infrastructure. With bring your own cloud (BYOC) on AWS, now in private beta, compute, build artifacts, and data run inside your own AWS account and VPC, while Vercel operates the control plane on top. Apps and agents reach private backends and internal systems the same way anything else in your account does, and source code never leaves your CI.
BYOC gives engineering teams the Vercel workflow while security retains the network controls, audit evidence, and account ownership it already has.
What is available now
The current availability breakdown is as follows:
- Vercel Passport (public beta): IdP-backed access control for all deployments, defaulting to private.
- Vercel Connect (beta): Short-lived, per-task credentials for agent access to external systems.
- Enterprise Managed Users (private beta): Full identity lifecycle management via SAML SSO and Directory Sync.
- BYOC on AWS (private beta): Workloads run inside the customer's AWS account and VPC.
The platform controls are framework-agnostic; Passport, Connect, and Enterprise Managed Users apply to anything deployed to Vercel, existing or new. Identity support includes Okta and any SAML or OIDC provider, while connect integrations cover OAuth- or API-reachable services. The goal is to make the safe path the default so that internal tools and agents can move from prototype to production without a security review becoming the bottleneck.
