Rapid Booking Site Build: FavCRM and AI for UI

A detailed playbook reveals how a booking site, traditionally a two-week project, was deployed in a single afternoon. This method leverages a headless CRM and AI-driven UI generation for accelerated development.

A FavCRM blog post detailed how a booking site, previously estimated to require two weeks of development, was built in a single afternoon. This acceleration relied on a combination of a headless CRM backend and an AI agent for UI generation, demonstrating a rapid deployment model for specific client needs.

The scenario involved an agency building a booking site for a small clinic with three services and a single practitioner, requiring online booking with a deposit. The core shift from a multi-week timeline to an afternoon was attributed to offloading complex backend logic to FavCRM and frontend generation to an AI agent.

Register the Workspace

The initial step involved registering the clinic's FavCRM workspace, a process the founder claimed took approximately five minutes. This was executed via the favcrm command-line interface (CLI) rather than a web dashboard. The process involved two commands: favcrm signup request with email and organization name, followed by favcrm signup verify using a request ID and a 6-digit code. The verification step outputs a fav_mcp_* API key. This key is critical for subsequent operations and must be stored securely, typically as an environment variable, never committed directly into a repository.

Configure Services and Availability

Configuring the clinic's services and staff availability was reported to take around 30 minutes. This involved using the favcrm tool call command, often after inspecting the schema with favcrm tool describe. For instance, services like "New Patient Exam" (45 minutes, $80.00) and "Cleaning" (30 minutes, $60.00) were created through direct CLI calls. Staff availability, such as a Monday schedule from 09:00 to 17:00, was also set using a similar favcrm tool call set_staff_availability command. At this stage, the backend logic for services, hours, and an availability engine capable of handling booking clashes was fully configured without manual schema definition.

Implement One Server Route

A crucial security measure involved proxying FavCRM API calls through a single server route, preventing the API key from being exposed in the browser. This step was estimated to take 30 minutes. The example provided a Next.js route handler (app/api/booking/route.js) that interacts with FavCRM's MCP endpoint, which communicates via JSON-RPC over HTTP. The handler included a callTool asynchronous function designed to proxy two primary operations: reading available booking slots and creating a new booking. This centralized route ensures that all sensitive backend interactions are handled server-side.

Generate the Booking UI with AI

With the backend configured and the server route established, the next step involved using an AI agent to build the booking user interface. The source states, "Let the agent build the booking UI against that route." While specific time estimates for this step were not provided, the overall

Pull quote: “”