.jpg)
According to a BCG survey, 64% of business leaders have already begun implementing generative AI to reshape their internal organizations, using generative AI tools to save five hours of work each week, allowing them to focus on executing more tasks or handling strategic tasks. However, as generative AI brings more convenience to daily life, it also introduces challenges such as AI hallucinations, data privacy and protection, and the training of custom models. In a forum hosted by Tech Orange during the "AI Innovation Day: 2025 AI Innovation Day," Netron Information Technology's Senior Technical Manager Zhou Haowen delved into how enterprises can leverage flexible, secure AI knowledge management systems to accelerate industry innovation through generative AI, while overcoming the challenges they face in its use.
"When individuals widely use generative AI, why are enterprises relatively hesitant?" Zhou Haowen analyzed three key factors. First, when businesses use ChatGPT, they often encounter AI hallucinations, receiving biased, incorrect, or outdated information. Second, enterprises using ChatGPT find that Data Control, including the "Improve the model for everyone" function, means the data and entire conversation shared with ChatGPT may become part of its model optimization data, raising concerns about data privacy and security. Third, for reasons of security and the need to use the latest data, enterprises may want to train and fine-tune their own models. However, this requires significant financial investment and time to maintain and keep the model up to date. Even if the model is industry-leading, it still may not achieve the most accurate or efficient results.
NAVI Enterprise AI Knowledge Management System Protects the Use of Sensitive Data
To help enterprises overcome challenges related to generative AI, such as hallucinations, data privacy, and protection, Netron Information Technology has designed a "Generative AI Knowledge Management" solution tailored for businesses. This solution supports various data sources and integrates company IDs, categorizing data usage levels to protect sensitive information. Additionally, with the continued development of large language models, NAVI supports a wide range of advanced LLM models, including Claude 3.5, GPT-4, and Stable Diffusion, helping clients balance efficiency and cost. It also allows for easy budget management through LLM token quota control. "The technology we use is called RAG (Retrieval-Augmented Generation). You can think of it as providing a company with an expert or a highly experienced employee who might not know everything but can read hundreds of pages of documents in seconds and tell you related concepts or applications based on those files," explains Zhou Haowen.
How can enterprises apply generative AI knowledge management solutions? Zhou Haowen explains that in the healthcare field, Netron Information Technology can assist doctors by providing treatment or symptom information and answering patient queries in real-time. In the financial sector, the solution can automate responses to common customer inquiries, generate tips or best practices for managing customer accounts. In the education sector, it can provide personalized learning experiences based on student profiles and answer questions on academic subjects. In marketing, it can create ad copy for target ads and offer customer behavior insights through prompts. In human resources, it can respond to employee or employer questions in real-time and automate tasks like onboarding. In the tourism sector, it can generate content to provide travel and destination recommendations to customers, write travel content, and generate organic leads.
Innovative Application Scenarios in Healthcare, Finance, and Customer Service
"I have also discussed with many financial clients how AI can help accelerate their investment and financial analysis," said Zhou Haowen. Investment firms need to study the latest financial report data, which often includes charts and documents spanning several months for each report. They also need to frequently refer to more up-to-date websites, news, and social media information to develop investment strategies. However, when investment firms manually analyze data and then transfer it into their research Excel files and maintain the data, it usually takes several days. "Now, by using AI, we can assist enterprises with automation for analysis and report generation. Even if the documents are in English, Korean, or Japanese, they can be converted into the client's most familiar language, significantly shortening the entire analysis process from several days to just a few seconds. Additionally, in one or two hours of earnings calls, we can have AI automatically generate meeting summaries and compile investor questions, bringing significant convenience," Zhou Haowen added.
In the customer service field, Netron Information Technology also helps clients use AI to solve societal issues. Zhou Haowen gave an example of an emergency call at a hospital's emergency department. When a call comes in, emergency responders need to record all details, provide immediate rescue SOPs, and constantly confirm the patient's condition before logging the complete voice message into the system, all under immense pressure and a large administrative workload. Through AI assistance, the case details, including the type of case, the caller, the location, the case status, and the suggested steps for handling, can be automatically compiled from the call record. Zhou Haowen emphasized that Netron Information Technology aims to use its AI knowledge management platform to integrate with various business scenarios and provide more innovative applications, continuing to drive business innovation.
