The Endpoint Solutions team at UCLA Digital & Technology Solutions (DTS) is exploring the future of IT support with a bold experiment: using OpenAI's GPT technology to improve how we manage service tickets. The result? Faster triage, smarter workflows, and a powerful proof of concept for AI-powered productivity.
The Challenge
Manual ticket routing and repetitive templates were slowing down the Endpoint Solutions team’s response times. Common IT support tasks, like onboarding new email accounts or processing service requests, required staff to copy, paste, and interpret details across multiple systems. These bottlenecks created inefficiencies and limited the team's capacity for strategic work.
The Solution
The team in charge of this pilot, led an initiative to transform support workflows using generative AI. The team developed over 15 custom GPTs to streamline tasks in ServiceNow, where much of their work is concentrated. Without API access, they used a simple copy-paste method to input ticket text into custom GPTs and receive actionable outputs.
Key tools included:
- Triage Assistant: Suggests the appropriate technician and routing group by parsing raw ticket data and applying internal rules
- EM Account Template Assistant: Generates onboarding email templates from ticket content
- FedEx Label Request Generator: Automates Admin Services shipping label creation
- ServiceNow Ticket Assistant: Summarizes ticket details and drafts public comments and suggested next steps
All tools simulate the reasoning of human analysts and help ensure consistency across workflows, especially useful for onboarding and training new team members.
The Impact
This pilot project delivered measurable improvements:
- Faster, more consistent ticket triage
- Shorter turnaround times for routine tasks
- Clearer communication and fewer stalled tickets
- Improved learning and collaboration within the team
Beyond automation, the process encouraged critical thinking about existing systems and opened new avenues for experimentation and peer learning. A key insight: GPTs work best for repeatable tasks where logic and context are predictable. Even when GPTs don’t provide perfect outputs, they help get the ball rolling and reduce manual overhead.
Looking Ahead
To address prompt size limitations and avoid information overload, the team plans to implement RAG (Retrieval-Augmented Generation). This architecture will allow GPTs to access only relevant documents from a knowledge base, enabling smarter, more context-aware responses. Future plans also include:
- API integrations with ServiceNow and Confluence to automate data flows
- RAG-enhanced triage bots for scalable, policy-aware decision-making
- Continued human oversight to ensure quality, security, and trust
Advice for Other Teams
“Just start experimenting,” says Clark Kringel, Senior Analyst on the Endpoint Solutions team. “Custom GPTs are easy to prototype. The key is to begin with structured, repeatable tasks and iterate from there.” Cross-team collaboration and shared learnings can accelerate impact and reveal unexpected opportunities.
Learn More
UCLA DTS is committed to harnessing the power of AI to improve how we work, support, and serve. Visit our Newsroom to explore more of our AI innovations or connect with us to get involved.
Learn more or explore collaboration opportunities with UCLA Digital & Technology Solutions.
