Incident: On August 10, the company officially released YongPt2.0. The company's big model has settled models in vertical fields such as finance, manpower, and supply chain, and can achieve a large-scale model training system with stronger professional capabilities.

Positioning as a “vertical model with deep understanding of enterprise services”: In the latest version of YongPT, the company launched three important products: “Smart Friends,” “Digital Smart Employees,” and “Intelligent Big Search,” which are based on intelligent (Agent), human-computer interaction (HCI), and knowledge enhancement (RAG) application frameworks, respectively. Among them, Zhiyou plays the role of corporate Copilot, and digitally connects smart employees with enterprise RPA to help employees get rid of complicated and repetitive tasks. Intelligent search opens up the enterprise's internal knowledge base, allows the business process to be accompanied by knowledge throughout the process, and allows experience to settle into knowledge that can be passed on.

Connect the general model with enterprise applications to achieve differentiated advantages: In terms of the overall architecture, the bottom layer of YongPt2.0 uses general models such as Baichuan and Tongyi Qianwen, while between the general model and the enterprise application, users form the middle three layers to meet the needs of the enterprise service sector. The first layer is the YongPT2.0 model, which has been enhanced with 8 major professional capabilities, including the ability to generate PPT analysis reports, intelligent contract review capabilities, and contract generation capabilities. The second layer is a large model platform that includes capabilities and tools such as plug-ins, prompt engineering, fine tuning, etc., which can expand the storage, application, and scheduling execution capabilities of large models, and combine knowledge and leading practices in the fields of finance, taxation, human resources, etc., to expand the professional capabilities of large models, thus forming a systematic enterprise service model. The third layer is three general-purpose products (Smart Friends, Digital Smart Employees, and Intelligent Big Search) launched based on the three frameworks of Agent, RAG, and HCI.

24Q2 performance picked up, and we are optimistic about the implementation of full-year results. Earlier, the company officially disclosed its semi-annual performance forecast on July 9. In the first half of 2024, in accordance with the annual management policy of “seizing market opportunities, leveraging user advantages, and comprehensive efficient management”, the company steadily and solidly promoted various tasks to achieve steady revenue growth, improve management, reduce personnel size, and control costs and expenses. The first half of the year is expected to achieve operating income of 3.71 billion yuan to 3.88 billion yuan, an increase of 10% to 15% over the previous year; the net loss attributable to owners of the parent company is expected to be 0.75 billion yuan to 0.884 billion yuan in the first half of 2024; the net loss attributable to owners of the parent company after deducting non-recurring profit and loss in the first half of 2024 is 0.745 billion yuan to 0.88 billion yuan, a year-on-year decrease of 0.06 billion yuan to 0.195 billion yuan

Profit forecasting, valuation and rating: Considering that the company promoted organizational model upgrades in 2023, which had a significant phased impact on contract signing, implementation and delivery, and revenue recognition, the 24-25 net profit forecast was lowered to 0.138/0.399 billion yuan (adjustment range -82.2%/-69.6%), and the net profit forecast for 26 years was 0.489 billion yuan; PE for 2024-2026 was 209x/73x/59x; considering that the company model has already been upgraded It was basically completed in 2023, laying a good foundation for the company's long-term customer operation and performance development. It is optimistic about the company's future development as a leading enterprise application software service provider in China, and maintains a “buy” rating.

Risk warning: Product promotion fell short of expectations; market competition intensified; the effect of the big model YongPT fell short of expectations.