Recording and Highlights: Hear the latest on Box AI integrations with ServiceNow & Salesforce

Thanks to ​@daria and for yesterday’s overview of Box AI integrations for Salesforce and ServiceNow and ​@nread Box for engaging in the Q&A. Also thanks to our guests who joined and participated in the discussion. 

Introduction
The roundtable covered Box AI integrations, with an emphasis on ServiceNow and Salesforce. The session began with polls to gauge the audience's current usage of general AI tools and specific Box AI features, revealing a wide range of adoption from initial exploration to full integration.

Main Topics
 

What is Box AI?
​@daria  provided a brief overview of Box AI, designed to accelerate workflows by extracting insights from the breadth of unstructured data stored in Box.

The Box MCP Server
​@daria  introduced the Box MCP (Model Context Protocol) Server, the technological backbone for these integrations. She explained that the MCP Server acts as a secure bridge, enabling leading AI agent platforms like Salesforce Agent Force and ServiceNow AI Agent Fabric to connect to Box data without exposing raw file content. This ensures that all interactions respect Box's security and permissions. A brief demo illustrated how an integration with Claude, powered by the MCP server, could analyze financial statements from Box and generate a new proposal without the user ever leaving the Claude interface.

Box AI Integration with ServiceNow
 

The integration allows users to perform content searches, extract key data, and generate summaries from Box files directly within the Now Assist chat. A demo showed a user asking for "confidentiality obligations" from a specific vendor contract, and the Box AI agent successfully found the relevant document in Box, extracted the precise terms, and presented them in the chat.

Box AI Integrations with Salesforce
The discussion then shifted to Salesforce, covering two major updates:
1. Box AI Data Extraction for Salesforce: ​@daria  announced this new feature, which is now available in the Box for Salesforce managed package. It is designed to eliminate manual data entry by automatically extracting information from documents in Box to create new or update existing Salesforce records. The feature supports both standard and custom objects, includes a review and validation step for users, and can handle complex data structures like invoices with line items.

2. Box for Agent Force Integration: The team showcased an end-to-end workflow using Box AI within Salesforce Agent Force. The demonstration followed a sales representative who used the conversational AI to:
- Summarize an entire account history by pulling data from multiple documents in Box.
- Automatically update the corresponding Salesforce record fields with the newly summarized information.
- Generate a new proposal document using Box DocGen.
- Route the newly created proposal for e-signature using Box Sign.

This entire multi-step process was completed within the single Agent Force chat interface, highlighting significant potential gains in productivity.

Q&A
Key questions covered Box DocGen, how it handles file creation from Salesforce, and where to find technical resources and demos. Resources include Box's public Medium site and the developer advocacy team will be creating more resources.

Conclusion
​@daria  and ​@nread Box  reiterated their willingness to be a resource for the community, encouraging users to provide feedback and share their use cases to help shape the future of the product, in the user group.