Back office ops · Production

NTT DATA builds GenAI POC to classify and prioritize work orders for international infrastructure company

The problem

A leading international infrastructure company manually processed roughly 1,500 work orders per day using 70 employees, creating significant opportunity for error and inconsistency in categorization across more than 500,000 annual maintenance requests.

Workflow diagram · grounded in source
1
Work order request submitted
trigger
“The team uses the information and notes that operators and/or technicians enter in the system”
2
Knowledge base loaded into LLM
integration
“The client had well-documented policies, procedures, requirements and a comprehensive list of over 160 work order categories. Our team used this information to prompt the LLM on the intricacies of accurately classifying incoming requests”
3
LLM classifies work order
ai_action
“What expertise is required to address the resident's problem (classification)? What is the urgency level for this request (time prioritization)? Are there any special-handling requirements (for example, the resident wasn't home, a return…”
4
LLM explains classification reasoning
output
“the LLM not only classifies each work order before moving on to the next but explains why each one was categorized the way it was. This transparency into the "why" provides a valuable audit trail that enhances accountability and provides…”
5
Policy updates refine LLM
feedback_loop
“the client will use a custom application to adjust its policies and add clarifications for the LLM, so the process will continue to improve over time”
Reported outcome

The GenAI POC demonstrated the capability to classify work orders more quickly, accurately, and consistently than the manual approach, with the LLM providing an audit trail explaining each classification; the company plans full deployment once the LLM consistently demonstrates its usefulness.

Reported metrics
Annual maintenance requests500,000-plus
Work orders processed per day (manual baseline)about 1,500
Employees manually processing work orders70
Work order categoriesover 160
Show all 6 reported metrics
annual maintenance requests500,000-plus
work orders processed per day (manual baseline)about 1,500
employees manually processing work orders70
work order categoriesover 160
POC development timeunder two weeks
classification speed, accuracy, and consistencymore quickly, accurately and consistently classify work orders
Reported stack
large language model (LLM)
Source
https://us.nttdata.com/en/blog/2024/june/lessons-learned-from-a-genai-proof-of-concept
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The GenAI POC demonstrated the capability to classify work orders more quickly, accurately, and consistently than the manual approach, with the LLM providing an audit trail explaining each classification; the company…

What tools did this team use?

large language model (LLM).

What results were reported?

Annual maintenance requests: 500,000-plus; Work orders processed per day (manual baseline): about 1,500; Employees manually processing work orders: 70; Work order categories: over 160 (source-reported, not independently verified).

How is this back office ops AI workflow structured?

Work order request submitted → Knowledge base loaded into LLM → LLM classifies work order → LLM explains classification reasoning → Policy updates refine LLM.

Related back office ops cases