{"id":11059,"date":"2026-01-14T17:45:57","date_gmt":"2026-01-14T09:45:57","guid":{"rendered":"https:\/\/www.salesforce.com\/?p=11059"},"modified":"2026-01-14T17:45:59","modified_gmt":"2026-01-14T09:45:59","slug":"ai-agents-self-service-tickets","status":"publish","type":"post","link":"https:\/\/www.salesforce.com\/ap\/blog\/ai-agents-self-service-tickets\/","title":{"rendered":"How Our AI Agents Are Eliminating 70% of Employee Self-Service IT Tickets"},"content":{"rendered":"\n

At a company the size of Salesforce, a “simple” employee change can be anything but, demanding flawless orchestration of data across 100+ internal and external systems, including HR, payroll, benefits, identity management and dozens of vendor applications. <\/p>\n\n\n\n

That\u2019s why we turned to our own technology<\/a> to streamline processes like our onboarding using MuleSoft<\/a> and Agentforce 360<\/a>. Now we\u2019re on track to cut 70% of employee self-service IT tickets.\u00a0<\/p>\n\n\n\n

Before using MuleSoft and Agentforce 360 for onboarding when an employee changed roles and noticed data discrepancy in different systems, it kicked off a long process with a lot of manual tickets and follow ups.<\/p>\n\n\n\n

A single mismatch, like a failed permissions sync due to an old department code, would mean a locked-out employee. And that locked-out employee\u2019s first stop? Filing an IT help-desk ticket.<\/p>\n\n\n\n

This wasn’t just an inconvenience; it was a massive drain. We were burning approximately 2,000 high-value IT work-hours each year on “swivel-chair” data cleanup: an engineer manually cross-referencing multiple systems to find one bad entry. It was slow, expensive, and not a great  experience for our employees.<\/p>\n\n\n\n

We knew we had to do better. Here\u2019s how we made this tedious process an efficient one, improving the employee experience<\/a> so we could better serve our customers.<\/p>\n\n\n\n\n\n\n\n\n\n\n

The solution: From reactive tickets to proactive AI<\/strong><\/h2>\n\n\n\n

We tackled this with a two-part strategy.<\/p>\n\n\n\n

Part 1: Building the foundation with MuleSoft<\/strong><\/h3>\n\n\n\n

First, we had to build a single source of truth for all 90+ systems. Our integration team used MuleSoft to build a network of APIs that securely unlock data from every core employee application.<\/p>\n\n\n\n

This created a unified “Employee 360”<\/strong> profile, allowing all systems to subscribe to a single, trusted stream of data. This alone was a huge step forward, but it was still reactive. If a downstream system failed to “get” the message, the data would cause discrepancies in downstream systems.<\/p>\n\n\n\n

Part 2: The game-changer: Data 36O and Agentforce<\/strong><\/h3>\n\n\n\n

This is where we flipped the model. Instead of waiting for an employee to file a ticket, we used Agentforce<\/a> to support the employees for correcting these discrepancies as a self-service capability.<\/p>\n\n\n\n

To begin, we leveraged Data 36O<\/strong> as the centralised data fabric for our Agentforce agent in retrieving the context. This was essential for storing data (like integration documentation, field mapping information, FAQs and support logs) and providing system specific context to the Agent.\u00a0<\/p>\n\n\n\n

Agentforce constantly supports our employees through our employee internal self-service portal (BaseCamp) and responds to all the user queries related to their profile information discrepancies. Here\u2019s what it does:<\/p>\n\n\n\n

    \n
  1. Identifies a mismatch:<\/strong> It detects when an employee’s status in one system is different from another system.<\/li>\n\n\n\n
  2. Understands the “why”:<\/strong> It doesn’t just find the error; it analyses it. It knows the difference between a simple processing lag and a genuine data conflict.<\/li>\n\n\n\n
  3. Takes autonomous action:<\/strong> Agentforce doesn’t just create an alert for a human. It invokes a MuleSoft API and autonomously fixes the data discrepancy on the spot.<\/strong><\/li>\n<\/ol>\n\n\n\n

    For example, if a role change fails to sync, Agentforce resolves the issue by using a simple natural language prompt from the user. No human, no ticket, no wait time for 70% of the issues.<\/p>\n\n\n\n

    The results: What this actually did for our business<\/strong><\/h2>\n\n\n\n

    By shifting from a reactive “human-in-the-loop” model to a proactive, “AI-in-the-loop first\u201d one, the impact was immediate.<\/p>\n\n\n\n