Page 327 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

for scenario optimization, which previously took three days for one developer and 1-2 days for
one tester, has been reduced to just one person completing it within a single day, accelerating
product iteration and optimization.

Increase in User Satisfaction: Based on dimensions such as survey questionnaires, online 4.3 - 5G
feedback, and usage trends, highly satisfactory results have been achieved. The satisfaction
rate from survey questionnaires is 95%, and the online evaluation satisfaction rate from frontline
users is 99.1%.

Opportunity Mining: Through the "Diting" Intelligent Customer Service Agent, potential
business opportunities are identified online by analyzing customer profiles and combining them
with marketing strategies to make personalized recommendations. Real-time data analysis and
service solution optimization accurately match user needs with business processes, avoiding
resource misallocation. On average, 15,000 potential business opportunities are identified
daily, with approximately 10% being valid opportunities.

2�3 Future Work

• Further explore the automatic planning and intelligent execution of service solutions:
Existing service solutions rely on the experience of business personnel to complete and
generate. In the later stage, consider using thinking chains and reinforcement learning to
automatically complete or generate service solutions for corresponding scenarios based
on real call data to improve the application of AI.
• Expand cross-industry applications: The "Diting" customer service agent has
demonstrated strong cross-industry adaptability, and will continue to promote its
application in e-commerce, finance and other industries. Customize intelligent customer
service solutions that meet the characteristics of the industry with various companies to
help companies improve service quality and customer satisfaction.
• Ecological construction and cooperation: The "Diting" customer service agent will
actively build a cross-industry ecological cooperation system. By establishing a developer
community, opening the API interface and development documents of "Diting",
encouraging R&D personnel to carry out secondary development, and jointly promoting
the advancement of intelligent customer service technology and ecological prosperity.
• In the future, the "Diting" customer service agent will continue to expand its application
scope and market influence, providing intelligent customer service solutions for more
industries and helping companies achieve digital transformation and upgrading.
• Transition to Human-Machine Conversation Mode: Further upgrade "Diting" into a
voice-enabled intelligent digital employee that directly provides human-to-machine
conversation services. The AI can directly engage in conversations with customers, offering
multi-round AI interaction capabilities, enhancing contextual understanding, improving
the model's ability to soothe customers through conversation, recognizing requests for
repetition, and optimizing the interactive experience for customers using online voice
self-service. As of June 3, 2025, the voice-enabled intelligent digital employee has been
deployed in 18 high-frequency human-machine conversation scenarios, with the rate of
complex conversations being transferred to human agents kept below 10%.

3 Use Case Requirements

• REQ-01: Data Requirements: It is critical to ensure the privacy and security of the data
during model training process. The data is sourced from the Hubei Telecom Customer
Service Center and consists of voice data between customer service personnel and
incoming callers.

291