Pronto is a fictional, AI-driven food delivery platform that uses AIforce and the Salesforce Platform to serve their customers.
The sample app ships with the following:
- Custom data model (storefronts, orders, reviews, menus...) along with the related sample records
- Storefront Explorer Lightning page
- LWR Site for customer support
- Agentforce Agents with their actions:
- Customer Service Assistant agent (browse storefronts, manage orders...)
- Employee Assistant agent (browse storefronts, view orders, view reviews...)
- Custom hosted MCP server with
- tools for order management and storefront browsing
- Headless Experience Layer (HXL) widgets for rendering a rich UI in third party agents (order status, storefront profile, storefront menu) and Agentforce actions (customer reviews, storefront profile, storefront menu)
The Pronto app is set up via an installation script that requires the following:
- Salesforce CLI's latest version. Here's how you can install or update it.
- Node.js LTS (v22 or above)
- if running on macOS or Unix: jq
- if running on Windows: PowerShell
If installing the app on a scratch org, you will need a Dev Hub Org configured with the Salesforce CLI. You can sign up for a free Developer Edition Org for this.
If installing the app on a Sandbox, a Developer Edition org or a Trailhead Playground, enable all of the following features. Failure to do so will cause the installation to fail.
- Go to Setup > User Interface and click Enable Custom Addresses
- Go to Setup > State and Country/Territory Picklists and click Enable State and Country/Territory Picklists
- Go to Setup > Digital Experiences > Settings and click Enable ExperienceBundle Metadata API
- Go to Setup > Omni-Channel Settings and enable Enable Omni-Channel
- Go to Setup > Agentforce Studio > Agentforce Agents and enable Agentforce
- Go to Setup > Messaging Settings and enable Messaging
- Go to Setup > Feature Settings > Headless Experience Layer Settings and enable Headless Experience Layer (Beta)
-
Clone the project from GitHub.
-
Open a terminal in the project folder.
-
If installing on a non-scratch org, authenticate to your org and set it as the default org for the project with the Salesforce CLI:
sf auth web login --alias pronto --set-default
-
Run the installation script from the project root folder:
macOS and Unix:
./bin/install.sh
Windows PowerShell:
bin\install.bat
-
Publish the
Customer_Service_AgentEmbedded Service Deployment:The installation script automatically opens the org on the Embedded Service Deployment setup page. If you close the page, you can find it in Salesforce Setup by searching for
embeddedin the Quick Find box and clicking Embedded Service Deployments.- Click Customer_Service_Agent.
- Click Publish.
The installation script activates the Pronto Employee MCP server and prints its URL. The URL contains your org's My Domain, for example:
https://api.salesforce.com/platform/mcp/v1/d/orgfarm-0b99678488/custom/ProntoEmployeeServer
You can also find the URL in Setup: enter MCP Servers in the Quick Find box and open Pronto Employee Server.
To call the server from an agent such as Claude or ChatGPT, create an External Client App for that agent and connect the agent as described in Connecting an MCP Client. Create one External Client App per agent, as the callback URL is specific to the agent.
The Pronto app relies on the following objects (non-relationship fields are hidden from this diagram):
erDiagram
CONTACT ||--o{ STOREFRONT__C : "is primary contact for"
CONTACT ||--o{ ORDER__C : "places"
CONTACT ||--o{ REVIEW__C : "writes"
STOREFRONT__C ||--o{ MENU__C : "offers"
STOREFRONT__C ||--o{ ORDER__C : "receives"
STOREFRONT__C ||--o{ REVIEW__C : "receives"
STOREFRONT__C ||--o{ STOREFRONT_HOURS_OF_OPERATION__C : "has"
MENU__C ||--o{ MENU_ITEM__C : "contains"
ORDER__C ||--o{ ORDER_ITEM__C : "contains"
MENU_ITEM__C ||--o{ ORDER_ITEM__C : "ordered as"
CONTACT {
id Id PK
}
STOREFRONT__C {
id Id PK
lookup Account__c FK
lookup Primary_Contact__c FK
}
MENU__C {
id Id PK
masterdetail Storefront__c FK
}
MENU_ITEM__C {
id Id PK
lookup Menu__c FK
}
ORDER__C {
id Id PK
lookup Customer__c FK
masterdetail Storefront__c FK
}
ORDER_ITEM__C {
id Id PK
masterdetail Order__c FK
lookup Menu_Item__c FK
}
REVIEW__C {
id Id PK
masterdetail Storefront__c FK
lookup Customer__c FK
}
STOREFRONT_HOURS_OF_OPERATION__C {
id Id PK
masterdetail Storefront__c FK
}
