Session Overview

Sam researched third-party AI tools that can help reconfigure and enhance his Salesforce/Litify org. The session started broad, exploring enterprise options, then narrowed based on Sam's actual needs: tools that can read AND write to Salesforce, are easy to set up, and don't require advanced developer skills. Sam is the one who does all technical implementation work, not Nathaniel (the Salesforce admin).

The session evolved from "what AI tools exist for Salesforce" into a concrete three-tool implementation plan.

What Was Accomplished

Research Phase

Researched dozens of third-party AI tools for Salesforce across these categories:

• Admin automation and configuration (Clientell, Sweep, Cirra AI)
• DevOps and deployment (Gearset, Copado)
• Data quality (Plauti, DataGroomr, Cloudingo, DemandTools)
• Change impact analysis (Panaya)
• Workflow orchestration (Zapier)
• MCP connectors (GPTfy, Adzviser, CData, Cirra AI, Salesforce DX)
• CRM-specific AI tools (Arcade, Cotera, Apollo)

MCP Education

Sam asked what MCP (Model Context Protocol) means and how it works. Covered:

• MCP is an open standard created by Anthropic (now supported by OpenAI, Google, Microsoft)
• Analogy: USB-C for AI. One universal protocol for AI apps to connect to external data sources
• Three pieces: Host (Claude Desktop/Cursor), MCP Client (built into host), MCP Server (connects to Salesforce)
• Data flow: User asks question > Claude recognizes it needs SF data > MCP client sends request > MCP server queries Salesforce > data comes back > Claude answers
• Salesforce DX MCP Server has 60+ tools organized into toolsets: orgs, metadata, data, users, testing