API Workflows

API Workflows let your AI receptionist make real-time API calls during a live conversation. When a caller asks about something that requires a data lookup -- like pricing, order status, or appointment availability -- the AI can fetch that information from your systems on the spot and relay it to the caller.

Video walkthrough

Prerequisites

  • A Frontdesk AI account with an active receptionist
  • An API endpoint your receptionist should call (your own backend, a third-party API, etc.)

How it works

You define a workflow with:

  1. A scenario -- the situation that triggers the API call (e.g., "Whenever the caller asks anything about the MacBook")
  2. An API endpoint -- the URL, method, headers, and body
  3. Response mapping -- which parts of the API response to extract and what to tell the caller

When the AI detects that a caller's request matches your scenario, it fires the API call, extracts the data you specified, and uses it in the conversation.

Creating an API workflow

  1. Open your receptionist dashboard and navigate to the Overview page.
  2. Scroll to the API Workflows section.
  3. Click Add to open the workflow form.

Workflow fields

  • Name -- a descriptive label for the workflow (e.g., "Get MacBook Details")
  • Scenario -- describe when this workflow should trigger. Write this in plain language (e.g., "Trigger this action whenever the caller asks anything about the MacBook.")
  • URL -- the full API endpoint URL
  • Method -- GET, POST, PUT, or DELETE
  • Headers -- key-value pairs such as Authorization: Bearer <API_KEY> and Content-Type: application/json
  • Body -- the request body (for POST/PUT requests)
  • Response Data -- map JSON paths from the response to variable names and prompts that tell the AI how to use the data

Caller information extraction

You can configure the workflow to extract information from the caller during the conversation:

  • Extracted Call Info -- specific fields to collect from the caller (e.g., hard disk size, core preference), each with a name, type, and description
  • Extract from Context -- information to infer from the conversation history (e.g., whether the caller is an existing customer)

Additional options

  • Speech to Emit -- what the AI says while the API call is in progress (e.g., "Fetching MacBook details. Please hold on for a moment.")
  • Timeout -- how many seconds to wait before giving up on the API call
  • Repeatable -- whether the workflow can be triggered multiple times in a single call

Workflow types

Frontdesk supports two workflow types:

  • Scenario-based -- triggers when the caller's request matches a specific scenario
  • Knowledge Base Loader -- fetches data from an API and loads it into the AI's context as additional knowledge

Using pre-built templates

Instead of building a workflow from scratch, you can start from a template.

  1. Navigate to the API Workflow Templates page from your receptionist dashboard.
  2. Browse templates by category using the tabs at the top: CRM, Auto Repair, Beauty, Construction, Fitness, Food, HVAC, Legal, Real Estate, Retail, Technology, and more.
  3. Use the search bar to find templates by name.
  4. Click a template card to preview it, then apply it to your receptionist.

If you need an integration that is not listed, click the request integration option to submit a request to the Frontdesk team.

Managing workflows

Each saved workflow appears as a card in the API Workflows section showing:

  • The workflow name and scenario
  • An Edit button to modify the configuration
  • A Delete button to remove it
  • An Expand/Collapse toggle to view details
  • A Duplicate button to copy the workflow
  • An Error Logs button to view any recent failures
  • An enable/disable toggle to activate or deactivate the workflow

Workflow availability

You can configure when a workflow is active using the availability settings, restricting it to specific days and hours if needed.

Troubleshooting

  • API call timing out -- increase the timeout value or check that your API endpoint responds within the configured window.
  • Incorrect data returned -- verify that your JSON path expressions match the actual response structure. Test the API endpoint independently first.
  • Workflow not triggering -- make sure the scenario description clearly matches what callers are likely to say. The more specific the scenario, the more reliably the AI will activate it.
  • Error logs -- click the error logs icon on a workflow card to see recent failures and diagnose issues.