Kaiwu Core uses its "Exclusive Benefit AI" to break through the predicament of inventors and seeks seed round financing
In recent years, domestic awareness of intellectual property protection has continued to rise, and the annual number of patent applications has steadily climbed along with the growth of innovation investment. However, the supply-demand mismatch in inclusive patent services has persisted: traditional patent agency services have high barriers, with the cost of drafting a single invention patent application often ranging from several thousand to tens of thousands of yuan, and the process is lengthy, making it difficult to meet the efficient and low-cost needs of groups such as individual inventors, startup teams, and university faculty and students. Most AI intellectual property tools on the market are products wrapped around general large models, which commonly suffer from AI hallucination issues. Patents drafted using these tools are prone to problems such as inaccurate claim scopes and fabricated prior art, and the retrieval process often misses relevant documents, failing to satisfy the rigid demands of innovation entities for patent drafting and retrieval. Against this backdrop, the domestically developed AI-native patent tool "Zhuanqili AI" has entered the track with a self-developed patent-native large model trained on industry-private data, fundamentally avoiding AI hallucinations through generation rules that align with professional patent logic. The project is currently seeking seed round financing.
1. Supply-Demand Mismatch in Inclusive Patent Services Leaves AI IP Tools Trapped in the "Hallucination" Dilemma
The core barrier of patent services lies in their professional nature: practitioners must not only understand the innovative points of technical solutions, but also comply with patent authorization logic, and be familiar with the comparison rules of retrieval databases. This has kept the cost of traditional agency services persistently high, making it unaffordable for individual inventors, university students, and small and micro startup teams to pay drafting fees that often reach tens of thousands of yuan. As a result, these groups tend to turn to low-cost AI tools. However, most AI intellectual property products on the current market are simply fine-tuned from general large models, without professional corpus training in the patent domain, and do not grasp the core logic of patent drafting that "the protection scope is defined by technical features". The generated content often suffers from "hallucination" problems such as fabricated legal provisions, mismatched technical features, and missing core reference documents during retrieval. The resulting patents either fail to pass examination or have unreasonable protection scopes, which in turn brings potential risks to innovation entities.
As the number of patent applications by domestic individual and small-to-medium innovation entities increases year by year, there remains a gap in the mid-to-low end market for efficient and reliable AI patent tools. Products that both understand professional patent logic and can avoid AI hallucinations have become a rigid demand in the industry, and Zhuanqili AI has entered the market precisely targeting this unmet need.
2. Self-Developed Patent-Native Model Fundamentally Eliminates AI Hallucinations
Unlike most wrapped products, Zhuanqili AI adopts a self-developed patent-native large model, rather than superficial adaptation based on general large models. Its core generation logic is to "first extract the main technical features of the technical solution, and then the AI fills in the implementation details around these main features", which fully aligns with the professional rules of patent drafting. This avoids the vast majority of hallucination issues common in general models from the source of content generation, making the product's accuracy and practicality higher than the industry average.
From the launch of the project to the release of the official V3.0 version, the team has successively overcome two core challenges: insufficient efficiency of massive patent retrieval and AI-assisted 3D modeling adaptation. These related issues have been resolved in the V3.0 version, and the core technical solutions have not been made public for the time being due to trade secrets. After the product was launched, it has received special recommendations from *Huanggang Daily* and Huanggang News Network, and also won the third prize in the (Technical Algorithm Track) of the Xinghe Industry Application Innovation Award, which was selected under the joint guidance of the National Engineering Research Center for Deep Learning Technology and Application and the Alliance of Artificial Intelligence of China (AIIA). Its industry recognition continues to rise. At present, Zhuanqili AI has opened services through its official website www.zhuanqili.cn, which has been adapted for mobile devices. Mini-programs, mobile apps, and desktop PC products are all under planning.
3. Diverse-Background Team Drives Independent Development, Project Operates Sustainably While Seeking Seed Round Financing
Zhuanqili AI was founded by Lan Yang, who previously worked at Baidu and the Information Technology Center of AVIC, with a composite professional background covering large internet companies and industrial informatization. The project name "Zhuanqili" is derived from two ancient Chinese texts: *"The powerful people monopolize its use and seize exclusive benefits"* from *On Salt and Iron · Prohibiting Tillage*, and *"The Jiang family has enjoyed exclusive benefits for three generations"* from Liu Zongyuan's *The Snake Catcher's Story*. The ancient meaning refers to benefits being controlled by a few, but the project takes the opposite connotation — enabling ordinary inventors and small and micro innovators to also enjoy the exclusive benefits of their own innovations. In terms of commercialization, Zhuanqili AI's target users cover multiple groups such as individual inventors, startup teams, patent agencies, university teachers and students. Currently, it reaches precise users mainly through channels including Xiaohongshu, WeChat Official Accounts, WeChat Channels, Douyin, Weibo, Zhihu, and brand-built communities. Since its launch, it has received a large number of positive user feedbacks, with the most common comment being "This is a great product, keep going". At present, the project's funds come from team self-funding and product operating revenue, maintaining a stable self-sustaining operation state. This seed round financing will be mainly used for subsequent iterations of the native model, improvement of the product matrix, and market expansion through precise channels.
When talking about future plans, the Zhuanqili AI team stated that in the short term, they will focus on further improving patent retrieval efficiency, diagnosing problems in multiple stages of the patent application process in advance, and identifying patent conflict risks at an early stage. Later, they will gradually collect user feedback to continuously improve the overall service quality. The AI+intellectual property sector is a track that requires professional accumulation. As the saying goes, "Stay principled and pursue unique innovations; the road ahead is long", the team will first root in user demands and focus on refining the current product.