Addressing AI and compliance pain points to reconstruct GMP software, Yaodaixu is seeking angel round financing.
Hangzhou Yaodaixu Digital Technology Co., Ltd. (referred to as "Yaodaixu") is advancing its angel round financing, planning to raise RMB 12 million, transferring 20% of its equity, corresponding to a post-money valuation of RMB 60 million. Yaodaixu focuses on AI-driven intelligent management software for drug and medical device GMP compliance, aiming to upgrade GMP quality management from the old paradigm of "in-process recording, isolated modules, and loose logic" to the intelligent paradigm of "pre-warning, in-process assistance, and fully linked workflows".
The Overlooked GMP Data Silos and Untapped AI Implementation Gap
The GMP (Good Manufacturing Practices) system for pharmaceutical and medical device enterprises consists of more than 20 core modules including documents, training, change, deviation, CAPA, suppliers, LIMS, and audit, with massive data dependencies and process linkages between modules. However, most mainstream GMP management software on the market adopts the "digitalized process" design idea, with independently deployed modules and isolated data, forming widespread "data silos". Meanwhile, such products generally suffer from drawbacks including chaotic architecture, broken logic chains, counter-intuitive interaction, and cumbersome deployment, which can hardly fit the real business scenarios of enterprises.
In terms of AI application, the whole industry still stays at the L1 level of assisted Q&A. Most vendors only simply integrate general large model APIs to complete regulation retrieval and Q&A, without vertical domain fine-tuning, self-built knowledge base, access to underlying software logic, or real-time synchronization of regulatory updates. Mature products that can assist business operations and even support decision-making are still a blank in the current market.
Policy signals on the demand side are becoming clear. The "Implementation Opinions on 'Artificial Intelligence + Drug Regulation'" issued by NMPA (National Medical Products Administration) (No. 6 [2026] of the General Office of National Medical Products Administration), the "Implementation Plan for the Digital and Intelligent Transformation of the Pharmaceutical Industry (2025-2030)" jointly released by seven departments including the Ministry of Industry and Information Technology (No. 79 [2025] of the Ministry of Industry and Information Technology), and the new version of "Good Manufacturing Practices for Medical Devices" (No. 107 of 2025) all explicitly encourage digital and intelligent transformation.
Compliance by Design: Enabling AI to Understand the Underlying Logic of GMP
Yaodaixu's products consist of three parts: business modules, system settings, and AI agents. On the business side, more than 20 GMP modules are integrated into the same platform, driven by logic chains rather than function stacking, to realize natural data interconnection between modules. The AI agent is deeply embedded in the software in the form of a floating window, which can read the underlying operation logic and user operation data in a penetrating manner (read-only and no external upload), and deeply participate in business process control.
Its core capabilities are reflected in three aspects. First, it has built the GMP vertical knowledge base QBase, which structurally includes full regulations of NMPA, FDA, EMA, MHRA, PIC/S, WHO, more than 1000 compliance cases, and 483 deficiency items, with continuous updates. Second, AI fine-tuning based on QBase and GMP vertical corpus enables the model to understand quality management logic rather than output general-purpose replies. Third, the "Compliance by Design" architecture: the AI does not perform decision-making operations such as approval, submission, and release. All citations can be traced back via hyperlinks, the full decision path leaves complete traces, and the model version is bound to the generated content, which meets the requirements of EMA Annex 22 and FDA credibility framework at the architecture level rather than the patch level.
This design directly responds to the core contradiction of GMP management that "100% certainty is required while AI is essentially probabilistic" — it holds the bottom line of compliance through the mechanism of "AI assistance, manual confirmation, and full-process traceability", while releasing efficiency value. According to optimistic estimates, compliance deficiencies can be reduced by 80% to 95%, and the per capita efficiency of the quality team can be increased by 50% to 80%.
Adopt Modular Subscription to Enter the Early Clinical Stage and Target a 3.8 Billion RMB Market
In terms of business model, Yaodaixu takes the modular subscription system as the core. The annual fee for a single module ranges from RMB 80,000 to 150,000 (for every 20 users). The basic version of the AI model is free of charge, and the full version has an annual fee of RMB 200,000. The company also provides supporting services including verification fee, customized development, and integrated "Software + AI + Consulting" services. The customer acquisition path adopts the "accompanying growth" strategy: in the early stage, it establishes trust with early-stage clinical innovative drug customers through low-price trial and supporting GMP consulting, in the middle stage it promotes commercial promotion and channel cooperation, and in the later stage it explores overseas markets such as Southeast Asia and Europe.
Target users cover innovative pharmaceutical enterprises, traditional pharmaceutical enterprises, CDMOs and medical device enterprises. Among them, CDMOs have particularly urgent demands for digital and intelligent transformation because they need to undergo audits from multiple entrusting parties. Innovative pharmaceutical enterprises and traditional pharmaceutical enterprises, under the pressure of increasingly strict GMP compliance supervision, are the optimal customer resources for digital and intelligent software. According to public data calculation, there are about 19,000 potential domestic customers. Based on the median annual subscription of RMB 400,000 and 50% penetration rate by 2030, the domestic market size is about RMB 3.8 billion per year.
Interdisciplinary Team Ensures Perfect Product Delivery
The team has dual genes of "compliance + intelligence". CEO Pan Yi has more than 12 years of experience in CDMO and commercial pharmaceutical quality management (previously worked at Bio-Thera, JOINN, and Genscript ProBio). CTO Wang Yongguang has 13 years of experience in front-end and back-end development and AI model development (previously worked at Alibaba, ByteDance at P7 level). Product VP Kong Wenqi has 12 years of experience in GMP software implementation and verification (previously worked at WuXi Biologics, BeiGene, Novo Nordisk, and Asymchem). The three members form a closed capability loop in "understanding compliance, understanding AI and software, and understanding GMP software implementation". Their domain knowledge will be precipitated into the QBase knowledge base, fine-tuning corpus and implementation methodology, reducing reliance on individual experience.
At the current stage, the project is in the angel round. Core modules (document management, training, deviation/change/CAPA, etc.) are under development, the QBase knowledge base and AI agent are being debugged synchronously, and the supporting CSV (Computerized System Validation) acceleration solution can reduce the on-site customer verification cycle by nearly half. It is estimated that the revenue will grow from RMB 300,000 to RMB 72.74 million in the next five years, and EBITDA will turn positive in the fifth year.
In the window period when GMP software is evolving from "digitization" to "intelligence", whether Yaodaixu can carve out its own share from the existing market occupied by international manufacturers and domestic players with its differentiated positioning of AI-native and Compliance by Design depends on the verification progress of early customers and the continuous accumulation of the knowledge base.