The era of making money by relying on licenses has passed. Data service provider "Feidi Technology" intends to "earn service fees" for financial institutions | Project Report

Author｜Wu Sijin

Editor｜Deng Yongyi

In the past, financial investors obtained professional information such as stock recommendations and wealth management through financial institutions such as banks and securities companies to evaluate how to increase the value of assets. Now, relevant content is blooming everywhere... Media, Internet platforms, financial institutions, etc., are all producing content from different perspectives, influencing the minds of investors everywhere. When the boundaries of the above three begin to intersect and information becomes overloaded, what changes will occur in the demand for financial services by financial investors? And how should financial institutions respond?

36Kr recently came into contact with a financial data service provider, "Feidi Technology". Qiu Huihui, the founder and CEO of the company, believes that currently, the licenses of financial institutions are in excess, financial content is in excess, and after the era of large models arrives, IT resources are also in excess, but services are still scarce. The demands of investors for noise reduction and the needs to be served and accompanied have not yet been met. Financial institutions should shift from the seller model of earning commissions, transaction fees, and other "easy money" by relying on licenses to the buyer model of serving investors.

The buyer model mentioned by Qiu Huihui means that financial institutions not only provide investors with content information that can improve personal asset management education and cognition but also help users actively reduce the noise of overloaded information. In addition, it will give users a sense of companionship in the current fluctuating market and provide accurate financial services.

Therefore, the product idea of Feidi Technology is to start from the user's perspective and provide financial institutions with data products with content labels and an integrated solution of intelligent interaction AI to achieve this plan. The key to this solution is its self-developed "Feidi Financial Information Big Data Engine".

The Feidi Financial Information Big Data Engine obtains unstructured data through a massive amount of public network information, combines many self-developed expert models and knowledge bases of Feidi, extracts valuable data, and then generates data information with content labels according to the needs of different financial scenarios. Finally, it is packaged into data modules and content products that meet the needs of multi-scenario agile adaptation, such as position-holding companion content strategy data, hot topic operation content strategy data, and user growth and comfort content strategy data.

Figure: Flowchart of the capabilities of the Feidi Financial Information Big Data Engine

For the operation platform department of financial institutions, after the above data modules with content labels are connected to the enterprise client through API and combined with the user operation ideas of financial institutions, content that is tailored to individual users, different scenarios, different investment advisors, and different strategies can be launched on the front end to achieve functions such as Toutiao for securities, Weibo for funds, and push notifications for changes in self-selected stocks that users are concerned about.

By providing users with professional content after noise reduction, the open rate, subscription rate, and sharing rate can be improved, thereby promoting the efficiency and output-to-input ratio of the entire process from customer acquisition, retention, activation, to transaction conversion.

Figure: Application of Feidi products in high-frequency information scenarios of some financial institution clients

For the investment advisor department, after the above data products with content labels are packaged into AIGC tools, they can work with investment advisors, customer managers, brokers, and other staff in a human-machine collaboration mode to quickly respond to user questions from a professional content perspective, improve service efficiency and business development capabilities, and help enterprises build an Investment Advisor Copilot and an Investment Advisor Aigen.

For the middle platform department that hopes to achieve the goal of enterprise digital and intelligent transformation, Feidi data services and various data, algorithms, script models can be combined with the enterprise's own AI large model in the form of private deployment to build an enterprise content middle platform, providing topic inspiration and initial materials for enterprise operation personnel and content producers during the content creation process, thereby improving the timeliness of content production and the diversity of topics to serve the personalized needs of different users.

Figure: Business process of the institutional AIGC intelligent operation SaaS system

In Feidi's business model, it not only involves the production and distribution of content, the attributes of financial services, and user usage scenarios but also involves data mining engineering and AI application technology. It is itself a considerable challenge to productize the professional knowledge, experience, and service concepts in these different fields.

Qiu Huihui told 36Kr, "Feidi is helping institutions earn money from services. But so far, the main business of other similar professional data service providers is still content supply, while the professional adaptability of other engine platforms in the vertical field of finance is relatively low. Because there are no enterprises with very similar product functions to Feidi, many financial institutions have cooperated with Feidi through a single procurement process in 2024, which is very rare in the financial industry."

Feidi's ability to achieve productization is related to Qiu Huihui's past career background of "finance + content + entrepreneurship". Qiu Huihui once worked for "21st Century Business Herald", starting as a front-line reporter and later becoming an editorial board member & chief innovation officer and the person in charge of the important news section, with 20 years of experience in financial media and industrial reporting. In 2016, she started a business similar to the "Financial Information Big Data Engine" project as the founder of the AI News Laboratory within the newspaper and received investment from Sina. In January 2022, Qiu Huihui led the original team to start an independent business and established Feidi Technology. Currently, the Feidi team has more than 30 people, and the team members are mainly data research and technical R & D.

In Qiu Huihui's view, the biggest threshold of the Feidi model lies in domain Knowhow + data precipitation. Public network data + purchased third-party data + continuously expanding alternative data determine the width of the Feidi database. Only after application processing and deep modeling based on the financial service scenarios can it become Feidi's private data. These private data that meet the intelligent application needs of different scenarios determine the depth of the enterprise database, and this depth is also the moat that makes the Feidi model difficult to be replaced.

In addition, the Feidi team has been exploring the financial information big data engine product since 2016 and has currently accumulated a large amount of private data that has passed the market verification in terms of compliance and accuracy. This time-based first-mover advantage is also one of Feidi's competitiveness.

If AIGC wants to be applied in the field of finance, which has a high compliance threshold, low fault tolerance rate, and strong professionalism, the primary problem to be solved is to answer the accuracy and compliance of the content, and at the same time, eliminate the illusion problem of large model applications. However, to improve the accuracy rate, it is necessary to combine with the private database of a specific field for training to have a commercial basis.

In terms of the accuracy and compliance of the generated content, Qiu Huihui believes that only by combining AIGC with Feidi's private models and private data can the accuracy of the generated content be improved, the timeliness be enhanced, and the precision of scene adaptation be increased. At the same time, Feidi integrates human-machine collaboration into the entire process of information from the source, processing, production, to distribution, so as to ensure the efficiency of content generation while also ensuring the compliance of the generated content.

Currently, Feidi serves three types of financial institution clients: securities institutions, fund companies, and banks. According to the data from the statistics of the Fund Industry Association in the second quarter of 2023, among the top 15 securities institutions in the brokerage business of securities companies, 10 have already reached cooperation with Feidi Technology, including CITIC Securities, Huatai Securities, Guotai Junan Securities, China Galaxy Securities, CITIC Construction Investment Securities, China CICC Wealth Securities, China Merchants Securities, GF Securities, Guotou Securities, and Founder Securities. In addition, Yingmi Fund, which has a 20% market share in the fund investment advisor market, and HSBC Bank are also using Feidi products.

90% of the Feidi products currently used by the above enterprises are data products, and the charging model is to charge an annual subscription fee according to the data module. By the end of 2024, the annual income of Feidi data can reach the tens of millions level, and the data renewal rate after customer signing is 100%. The unit price of the existing data module products is 300,000 - 400,000 yuan, and some institutional customer orders exceed one million yuan, exceeding the product unit price and customer unit price of traditional licensed media and data providers that provide information and content services.

The other two product lines of Feidi are entering the quasi-commercialization stage. Among them, the AI product serving the investment advisor department is undergoing initial customer verification, and the AIGC intelligent operation data middle platform is about to enter the product verification stage.

It is reported that Feidi previously completed a 7 million yuan angel round of financing and is currently undergoing a Pre-A round of financing, with Hua Jun Capital serving as the financial advisor.