The Context
An industrial manufacturer with a modular industrial hardware system offered customers nearly 1,000 product combinations from a catalogue of ~200 parts, motors, and sensors. This flexibility created extreme complexity for the service team.
Service inquiries came from multiple sources: customer technicians, service partners, and the company’s own engineers. Overwhelmed by the variety, most defaulted to calling the service department. As a result, senior engineers—some of the company’s most expensive talent—were spending half their time fielding routine inquiries instead of solving critical problems.
The Challenge
The company had already invested in tools:
-
A system that pulled documentation by order number, but still forced users to navigate up to 30 different PDFs per case.
-
A mobile app with an AI translator to support international customers, but inquiries often lingered due to time zones.
-
Multiple chatbot projects (internal and external) that failed—answers were generic, hallucinated, or missed key details.
The root cause was clear: documented knowledge alone was not enough. Real expertise was buried in Jira and Confluence, where teams had been solving issues and sharing workarounds for years.
Our Approach
Arti designed a solution that combined both:
-
Structured documentation from technical drawings and troubleshooting tables.
-
Hidden knowledge extracted from Jira and Confluence—patterns of how issues were actually solved, what information was needed, and which component combinations caused friction.
From there:
-
Repetitive inquiries were automatically handled with accurate summaries and direct documentation links.
-
Complex cases were escalated to humans with pre-filled Jira issues, initial analysis, and complete context.
-
AI agents collected missing information proactively, reducing back-and-forth delays.
The Results
Even at pilot stage, the projected impact was clear:
-
65% faster resolution time for inquiries.
-
50% deflection rate on repetitive questions.
-
25% of senior engineer time freed for high-value tasks.
-
Service staff able to resolve 85% of requests independently.
The Impact
By refining both explicit and hidden knowledge, the service team could scale without overloading senior engineers. Customers received faster, more accurate support—even across languages and time zones. And the company built a foundation for continuous automation as its modular product line expanded.
