By: MICHAEL R. SISAK, AP News
Chen Yue, a requirements analyst at the Shanghai R&D Center of Tokyo Electron, can now complete SAP project requirements analysis in just 3 days, work that once took two full weeks. The transformative upgrade comes from the AI-SAP Intelligent Requirements Analysis System, developed by a team led by Fu Qiang, a leading expert in enterprise management informatization.
Photo Courtesy: Fu Qiang
“In the past, coordinating communication between Chinese and Japanese business departments alone took five days, and requirement deviations occurred frequently,” Chen noted. Powered by generative AI, the system automatically aligns with Japan’s JIS financial standards and China’s VAT invoicing rules, driving requirement deviation rates 15% lower than manual analysis. “The report passed review by the Japanese headquarters directly, with no revisions needed.”
Following two weeks of on-site investigations and data verification across 12 multinational enterprises in Shanghai, Tokyo, and Suzhou, plus two industry associations and the R&D team, the system’s performance has been rigorously validated:
It has been deployed in 49 enterprises, cutting the average 14day requirements cycle to just 72 hours, lowering deviation rates from 20% to 3.2%, and reducing upfront project costs by more than 30% per project.
“Requirement analysis failure is the leading cause of SAP project failure,” said Takahashi Kenta, Digital Transformation Director at Mitsubishi Chemical (China). In 2023, the firm’s East China plant SAP upgrade incurred an additional ¥800,000 in rectification costs and a one-month delay due to missing overseas warehouse data synchronization requirements.
The 2024 SAP Project Implementation White Paper by the China SAP User Association shows that 48% of SAP project delays stem from incomplete requirements, 35% of cost overruns are directly tied to requirements deviations, and the average project incurs ¥630,000 in losses as a result.
Lin Feng, SAP Consulting Director at Shanghai Zhilian Information Technology, explained that traditional analysis requires rounds of interviews, drafting, and cross-department confirmation, often a full week just for documentation. Gaps between business and technical teams distort requirements, while cross-border projects face added complexity from conflicting national compliance rules and business practices.
Wang Min, IT Manager at Ningbo Ruipu Automotive Electronics, shared a costly lesson: “One project missed EU GDPR compliance in manual analysis, forcing post-launch reworks that cost an extra month and ¥200,000.”
“The complexity of SAP systems has outgrown manual capability; technological innovation is inevitable for the industry,” said Zhang Jinsong, an informatization expert at the China Machinery Industry Federation.
At a technology lab in Pudong, Shanghai, Fu Qiang’s team demonstrated the system’s core workflow: upon uploading multilingual raw requirements, the system automatically completes requirement decomposition, industry matching, compliance verification, technical translation, and report generation with minimal human input.
“The core is our self-trained SAP-specific large model,” Fu explained. Over two years, his team compiled more than 1,000 successful project cases, 300+ industry standards, and compliance rules for over 50 countries and regions, building a knowledge base covering manufacturing, chemicals, electronics, and three other key sectors. Targeted tuning enables the model to precisely map business terms to technical indicators and capture implicit requirements.
In a live test, a vague document mentioning “inventory linkage” and “financial accounting” was parsed into 18 specific requirements within three minutes, matched to an automotive industry solution, and flagged for China VAT and EU GDPR compliance. “Manual work would take at least two days and still miss 3–5 critical items,” said Zhao Yu, R&D Director. The system also uses natural language interaction to clarify ambiguous demands.
To address cross-border challenges, the system integrates a multilingual engine and a compliance library to automatically resolve regional differences. In one Tokyo Electron ChinaJapan project, it reconciled “quarterly closing” vs. “monthly closing” practices with a dual-cycle accounting plan, avoiding later rework. Today, the system achieves 96.8% recognition accuracy for requirements and continues to improve through machine learning.
Data from Tokyo Electron’s Shanghai R&D Center shows the system cuts cross-border project requirements cycles from 14 days to 3, deviation rates from 18% to 2.7%, and per-project costs by ¥450,000. “Our recent East China plant upgrade went live two weeks early,” said Project Manager Li Ran.
As an early adopter, Ningbo Ruipu Automotive Electronics completed three SAP projects using the system. “Complex requirement analysis fell from two days to four hours, deviations from 17% to 3%, and total savings reached ¥320,000,” Wang confirmed.
In its Asia-Pacific supply chain SAP project, Mitsubishi Chemical (China) reduced requirements cycles from 21 days to 3, automatically resolved 15 compliance conflicts across China, Japan, and South Korea, and cut project costs by 38%. “Multilingual reports eliminated repeated reviews among global teams and drastically reduced communication costs,” Takahashi said.
Among 10 randomly sampled pilot enterprises, the system lifted requirements efficiency by an average of 78.6%, reduced rectification costs by ¥410,000 per project, cut cross-department communication costs by 60%, and improved cross-border compliance verification efficiency by 85%.
Mark Schmidt, Technical Consultant at SAP SE Asia Pacific, commented: “This system deeply integrates generative AI with SAP domain expertise, solving longstanding industry pain points. Its technical design and real-world results are world-class.”
The system’s launch has sent ripple effects through the industry. Eight international SAP service providers are in partnership talks to include it in their global implementation toolkits. Twenty-five provincial CIO associations in China have listed it as a “recommended tool for SAP projects,” with coverage expected to exceed 500 enterprises by 2025.
“This system will restructure the SAP consulting industry’s division of labor,” Lin Feng noted. Work that once required a five-person team can now be done by one person, freeing staff for high-value business optimization and tripling consulting efficiency.
Fu’s team is developing customized versions for retail, logistics, pharmaceuticals, and other sectors, with plans to add dynamic requirements tracking to monitor changes during implementation and autogenerate adjustment plans.
“We want to help more enterprises access efficient, accurate requirements analysis,” Fu said. This AI innovation, led by China-based enterprise management informatization expert Fu Qiang, is reshaping the early-stage implementation ecosystem of global SAP projects.
Media Contact
