Cursor has introduced Build Live Apps, a feature that lets users create and publish working web applications by describing what they want to an AI agent. The tool can generate the application, set up a database, configure sign-in, and host the finished product without requiring users to arrange each service separately.
Announced Nov. 5, the feature is available on paid Cursor plans that support cloud agents. That means individual users need at least the $30-per-month Premium plan, rather than the cheaper $8 Plus subscription.
According to Cursor’s announcement, the feature is designed for both developers and employees building internal tools. It can handle applications such as sales dashboards, registration systems, onboarding portals, and other tools that teams can share through a web link.
How Cursor Build Live Apps works
Users start by opening a cloud agent in Cursor and describing the application they want to create. The agent then handles much of the setup that would otherwise require several separate services.
The workflow includes:
- App development: Generates a web application from a natural-language request.
- Database setup: Creates a Postgres database when the application needs one.
- Authentication: Configures sign-in through Cursor SSO.
- Hosting: Provisions the infrastructure needed to run the application.
- Deployment: Publishes the app to a unique domain chosen by the user.
If an application needs to connect to an external system, such as a company’s data warehouse, Cursor asks for the necessary credentials during setup.
Users can build from a new repository or an existing project hosted on GitHub or Cursor Origin. GitLab integration is planned but isn’t available yet.
Unlike AI coding assistants that primarily generate or edit source code, Build Live Apps also handles deployment. Developers comparing different approaches can look at how Codex and Claude Code manage coding tasks, although those tools don’t offer the same integrated hosting workflow.
Publishing isn’t automatic. Users must explicitly request deployment, and Cursor runs security checks before making the application available.
What security controls are included?
Cursor runs two types of checks before an application goes live: deterministic checks that look for defined problems and AI-based reviews intended to catch additional issues.
The company specifically mentions detecting secrets embedded in application files and exposed credentials.
If a check fails, the agent identifies the problem and asks the user to fix it. Applications aren’t published automatically after the code is generated.
Team administrators also receive several controls:
- View applications and secrets created across their team.
- Add custom rules to the security review.
- Override a publication blocked by the review process.
- Unpublish applications when necessary.
These features give administrators some visibility into employee-created applications, but they don’t replace a full security review.
For example, Cursor hasn’t said whether its automated checks test every application for authorization flaws, such as one signed-in user being able to access another user’s records.
An app handling employee or customer information should therefore be tested with accounts that have different permissions before it’s shared more widely.
The ability to override a failed security check also means organizations need clear rules about who can approve exceptions.
Cursor’s release comes as AI coding platforms expand beyond generating code. Developers using tools such as Conductor have already been managing multiple coding agents, while Cursor is now combining development and hosting in a single workflow.
How sharing and access work
Once an application passes its security checks and the user publishes it, Cursor provides a live link.
On team plans, applications can be shared with specific people, groups, or the entire team. Access is controlled through Cursor’s sign-in system.
Individual subscribers can also make their applications publicly accessible.
Cursor supports API-key access for scripts and Grok Bot, allowing software to interact with a published application without someone manually opening it in a browser.
That flexibility introduces additional responsibilities. Before sharing an application, users should confirm who can open it, what information it exposes, and whether its external connections have appropriate permissions.
Team administrators should also review which employees can publish apps and whether the organization’s existing approval process covers tools created through AI agents.
Cursor Build Live Apps pricing and limitations
Build Live Apps is included with paid Cursor plans that have cloud agents enabled. Cursor doesn’t list a separate subscription specifically for the new feature.
Its current pricing page lists the following individual plans:
- Plus: $8 per month, without cloud agents.
- Premium: $30 per month, including cloud agents and code hosting.
- Super: $100 per month, with a larger usage allowance.
- Ultra: $200 per month, with the highest listed individual usage allowance.
Team plans start at $30 per user per month, while enterprise pricing depends on the organization’s requirements.
Cursor uses a shared AI usage allowance that refreshes weekly. Third-party AI models can incur separate usage charges.
The company hasn’t published a separate price schedule for the databases and hosting provisioned through Build Live Apps. Users planning applications with substantial storage or traffic should confirm any infrastructure limits or additional charges before deployment.
There are also unanswered questions about data residency, retention after an app is unpublished, and how ownership changes when the person who created an application leaves a team.
For now, Build Live Apps offers a faster route from an idea to a working web application, particularly for internal tools and prototypes. Teams handling sensitive information should start with a limited pilot, test access permissions, and confirm hosting and data-handling requirements before using it for production workloads.
Also read: Lovable’s growing enterprise adoption offers another example of how AI app builders are moving from quick prototypes into business software development.