In order to actively promote the "artificial intelligence" action and promote the in-depth integration of information technology, work and study, and talent training, Hua Yifeng completed the localization of DeepSeek on the company's intranet, providing platform support services for internal employees to carry out DeepSeek applications, and effectively improving work efficiency.
Through the localization deployment of DeepSeek large models, five intelligent core application systems have been successfully built: 1) intelligent product selection system, 2)AI tutors provide 7 x 24 hours of intelligent training for new employees, 3) research and development code AI quality inspectors, 4) cross-departmental data decision center, and 5) after-sales problem lightning response system. These five applications have successfully constructed Hua Yifeng's intelligent operation matrix.
·Input open parameter information to ask questions, selection efficiency increased by 200;
·SupportMulti-condition compound query, accuracy ratemore than 95%;
·24 hours to answer product/technical questions
·The training cycle is shortened by 40%, and the standard answer is pushed in real time.
·Automatic detection of embedded code logic errors
·Typical BUG identification accuracy rate is 91%,Save 50%Try Time
(Code plus comments to improve code readability)
(Identify code bugs, save manual inspection time)
·Intelligent analysis of production/sales/inventory data
·Automatic generation of visual reports, key indicator alerts
·Automatic classification of fault description (accuracy rate 95%)
·Real-time push solution library, response speed increased by 60%
Through the construction of localized knowledge base, the intelligent upgrade of enterprise knowledge assets has been realized, and nearly 20000 technical documents, product cases and solutions that have been deposited for many years have been systematically integrated and deeply activated, thus building a continuously evolving enterprise intelligence center.
In the next step, we will continue to promote the expansion and overall utilization of computing resources, and gradually develop them to customers to better serve customers.
Hua Yifeng Sensor has built a "data-model-application" closed-loop industry AI solution package by localizing the deployment of DeepSeek large models, and promotes the intelligent upgrade of the sensor field with the dual barriers of data assets and AI algorithms.
