Apps your whole team builds together, not just one person
Describe it in plain words, invite your team to refine it, and publish it to everyone when it’s ready.
Everything you can do with Apps.
Code Stay close to the code.
Data Build around your own data.
KarenAI features Put AI inside the workflow.
Publish & share Get it into your team’s hands.
Claude Sonnet
GPT-5
Gemini FlashModels Your choice of intelligence.
Connections Connected to your company.
Access & controls Access, on your terms.
Build: Start with the way you work.
Start with the way you work.
Build a useful first version, then shape every detail.
- Build through conversation: Describe a web app and have AI create or modify its interface, server routes, logic, and database-backed behavior.
- Ask and Agent modes: Discuss an idea or move into implementation with tools suited to each builder mode.
- App templates: Start from a configured app template and make it your own.
- Reusable components: Find, inspect, and copy component templates in the builder.
- Company branding: Apply your organization’s colors, fonts, and logos. Import a brand when the brand service is configured.
- AI-generated app icons: Generate or regenerate an icon for your app. Requires image generation setup.
- Live preview: See the running app beside its build conversation and navigate its detected pages.
- App conversation history: Keep separate build conversations and revisit earlier work.
Stay close to the code.
Inspect, refine, and recover your implementation as you build.
- Source code editor: Browse, inspect, and directly edit application files.
- Package installation: Add dependencies and inspect package files within the supported runtime.
- Code exploration: Delegate investigation of an existing app without changing it.
- Build and type checking: Run build and type checks, then feed failures back into the builder.
- Error-assisted repair: Bring preview errors, deployment errors, and runtime logs into the build conversation to help diagnose and fix problems.
- Restore earlier code: Return to a recorded commit from a conversation checkpoint. Restoring code trims later conversation messages; it does not restore database contents.
- Clone code locally: Get a temporary authenticated Git clone URL to work with the code locally.
Build around your own data.
Keep the interface and the information it needs together.
- App database: Persist application data in a database provisioned for the app.
- Data and schema browser: Inspect app tables, contents, and schemas from the editor.
- External databases: Use connected PostgreSQL, MySQL, and Redshift data with connection-specific permissions.
- Organization databases: Create and query shared organization-owned PostgreSQL databases, separate from app-local storage.
- Saved queries: Define reusable data queries and call them with inputs.
- Business-system integrations: Build interfaces and actions around supported connected services and accounts.
- Custom APIs and OpenAPI: Work with configured API definitions and generate clients from OpenAPI descriptions.
- MCP tools in the builder: Discover and call permitted tools from connected MCP servers while building. Deployed app integration depends on the tool.
Put AI inside the workflow.
Give your app intelligence it can call when it needs it.
- Built-in AI endpoints: Request generated text or structured objects from app code. Requires a configured model service.
- Agent-powered features: Call linked agents for scores, evaluations, triage, or structured answers, with app and agent access in place.
- Callable app APIs: Inspect app routes and let agents or authorized clients use endpoints with their required authentication.
- API tokens: Create and manage personal access tokens with the scopes and expiration your API use requires.
- App background functions: Inspect and run deployed background functions from the editor. Requires an app provisioned with an Inngest account.
Get it into your team’s hands.
Publish, manage access, and understand how your app is used.
- Hosted publishing: Publish to a deployed Vybe URL with authenticated access.
- Deployment status and logs: See build and deployment results, inspect logs, and track unpublished changes.
- App access controls: Keep an app personal, share it with selected users, or open it to your organization.
- Page-level access: Grant access to specific paths and pages within an app.
- Guest page sharing: Invite an external recipient to a published page with an emailed access link. Requires organization sharing links to be enabled and the app to be published.
- Usage analytics: Track unique users, page views, top users, and usage trends over time.
- Database query logs: Inspect external database query activity from the editor.
- App lifecycle: Favorite, rename, archive, and restore apps as your workspace evolves.
Your choice of intelligence.
Use the right model for the work in front of you.
- Multi-model choice: Choose supported models from OpenAI, Anthropic, Google, and additional providers through the model gateway. Set reasoning effort where supported.
- Automatic routing: Let Auto balance cost and capability, with routing tiers and organization-level configuration. Available models depend on your workspace.
- Credits and usage: Manage subscriptions, credit purchases, and billing. Review usage breakdowns and trends, with supported limits and alerts.
Connected to your company.
Bring existing systems, data, and knowledge into the same workspace.
- Connected accounts: Find and connect supported tools, manage multiple accounts, and rename or disconnect them. Choose personal or organization visibility.
- Integration discovery and actions: Discover available connectors and their supported actions. Native providers include Granola, Intercom, Luma, Supabase Management, Zuora Billing, and Anthropic Managed Agents. Provider setup and available actions vary by workspace.
- Custom MCP servers: Connect server URLs, discover tools, and make permitted calls. Setup supports HTTP/SSE and no-auth, API-key, or OAuth options.
- Custom APIs: Register API documentation and authenticated access so AI can work with your own systems.
- Databases and saved queries: Connect PostgreSQL, MySQL, or Redshift, use read-only access and supported SSH tunnels, and reuse queries with inputs. Share organization-owned PostgreSQL data.
- Shared knowledge and branding: Make organization knowledge and reusable procedures available alongside personal context. Set company colors, fonts, and logos; brand import requires its connected service.
Access, on your terms.
Give people and AI the resources they need, with clear boundaries.
- Resource permissions: Grant supported full, read-only, disabled, or custom action access by resource. Permission options vary by connection.
- Human-granted access: When an agent needs a resource, it can request access for a person to grant. Existing permissions still govern subsequent actions.
- Workspaces and roles: Invite members and guests, assign roles, and switch between organizations.
- Single sign-on: Configure SAML or OIDC and domain enforcement for your organization. Requires organization SSO setup.
- Workspace navigation: Find work with search and the command palette, organize favorites and libraries, and use responsive layouts with light or dark themes.
Stop working with AI alone
2 min setup · No credit card required
