TL;DR
- MSP-specific AI must be trained on your ticket data and business context to make intelligent decisions, unlike generic tools like ChatGPT that lack domain knowledge
- AI augmentation reduces time to resolution by providing level one technicians with contextual information, remediation suggestions, and historical ticket patterns
- Building AI solutions in-house wastes resources; level one technician roles will largely disappear within 10 years as automation advances
Summary
Rick Mutzel from Omega Systems shares his 12-year MSP's journey implementing AI-powered automation for service desk operations. The discussion centers on the critical distinction between generic AI tools like ChatGPT and purpose-built AI trained on MSP-specific data and ticket repositories. Mutzel emphasizes that effective AI must understand the business context and historical ticket patterns to make intelligent routing and remediation decisions. The implementation focuses on augmenting level one technicians with contextual information, remediation suggestions, and previous ticket histories to improve time to resolution and customer satisfaction. He advises against building AI solutions in-house, predicting that level one technician roles will largely be automated within ten years, allowing MSPs to promote staff into higher-value positions. The flexibility of their chosen platform (Zophic) enables custom model training for non-ticket use cases including HR and security operations center support.
Chapters
0:00 - Introduction and Automation Needs
0:35 - Generic vs MSP-Specific AI
1:04 - Impact on Service Desk Operations
1:46 - Future of MSP Staffing
Key Quotes
0:42 "To be able to accurately action items, they need to be aware of our business needs and the service tickets that are coming in, they need to actually have a repository to be able to make intelligent decisions."
1:21 "Time to resolution, I think, is the key indicator. So being able to resolve the ticket correctly the first time in as quick of a manner as possible is definitely driving customer satisfaction ratings as a benefit of that as well."
1:38 "Don't try to develop it yourself. You will waste time, money and resources, and ultimately end up in the same situation."
