Integrating AI & Cloud for Financial Leap

In recent years, against the backdrop of declining interest rate centers, narrowing interest spreads, and economic transformation, the financial industry has abandoned the single pursuit of scale expansion and turned to seek high-quality development.

As the culminating work of the "five major articles," digital finance is not only the foundation and important cornerstone of financial development but also the core strategy for the financial industry to achieve high-quality growth, injecting new vitality and momentum into the financial industry.

As a technology and internet company, Tencent has been continuously investing in R&D strategically for many years, with a cumulative investment of over 300 billion yuan from 2018 to the present.

Based on capabilities such as cloud, databases, operating systems, security, big data, AI, and more, Tencent has been continuously creating a more "user-friendly" full-stack independent innovation product system and fully adapting to the mainstream domestic software and hardware ecosystem.

Advertisement

While building a new generation of digital finance foundations for the financial industry, Tencent actively explores the application of AI large models to help the financial industry achieve a new qualitative leap.

With the help of digital technologies such as cloud computing, big data, artificial intelligence, and blockchain, the financial industry can break data silos, continuously improve financial service efficiency, reduce service costs, expand the boundaries of financial services, enhance financial risk management capabilities, and comprehensively improve the quality and efficiency of financial services for the real economy.

As the lifeblood of the national economy, achieving comprehensive independent innovation is an important guarantee for financial security and long-term development.

Since 2019, the independent innovation work in the financial industry has made significant progress from batch pilot programs to large-scale applications.

Financial institutions have gradually moved from peripheral systems to core business systems, with the number of transformations increasing year by year.

At the same time, the industry side has continuously increased the supply of technological innovation and products, forming a complete ecosystem from basic software and hardware to core application systems, and continuously breaking through in aspects such as product maturity and technical autonomy.

It can be said that domestic independent innovation products are maturing from "usable" to "user-friendly."

However, for independent innovation products to be "user-friendly," it means continuously challenging performance limits, improving the cost-effectiveness and security of software, and promoting the perfection of the software and hardware industry chain and the prosperity of the upstream and downstream ecosystem, which poses higher requirements for all parties in the industry.

Technological breakthroughs require not only hardware breakthroughs but also software "climbing to the top."

For more than 20 years, Tencent has been continuously investing in R&D, continuously breaking through at the forefront of the software battlefield.

As of June this year, Tencent has more than 80,000 patent applications worldwide, supporting the continuous breakthrough of Tencent's integrated innovation products, achieving full self-research of product core code, turning software from a "black box" into a "white box," and accelerating the creation of a more "user-friendly" independent innovation software system.

On September 5th, at the 2024 Tencent Global Digital Ecosystem Conference, Tencent Cloud's full-stack integrated innovation software system was fully upgraded.

Tencent Cloud can provide financial institutions with basic software such as database TDSQL, operating system TencentOS, private cloud TCE, big data TBDS, and Tencent Cloud TI platform.

In terms of application software, it offers products such as Tencent Meeting, Enterprise WeChat, Tencent Docs, and Tencent Enjoy.

To date, Tencent Cloud has nearly a hundred products entering the working committee's hardware and software map, with industrial supply capabilities at the forefront of integrated innovation manufacturers; it has open-sourced more than 160 projects and has obtained nearly 1,400 mutual certification certificates, making it one of the most important manufacturers in the integrated innovation ecosystem.

In terms of technical foundations, Tencent's private cloud TCE has become the preferred private cloud software for the financial industry.

Through continuous evolution, Tencent's private cloud TCE has met requirements such as multi-core, cloud-edge integration, and consistent experience, helping more than a hundred financial institutions build private cloud platforms and manage millions of nodes.

Among them, the industry's largest "CCB Cloud" has been built, supporting more than 60,000 nodes, achieving unified management of multi-functional areas, multi-regions, and multi-chips.

At the same time, based on the advantages of TCE, Tencent and Shenzhen Stock Exchange have jointly created the industry cloud "Shenzhen Cloud" and are actively preparing for the financial group cloud to further help the financial industry innovate and improve quality and efficiency.

With the deepening of independent innovation in the financial industry's database, the increase in scenarios, systems, and equipment has become the norm for the industry, along with the increase in cost reduction and efficiency and operation and maintenance needs.

Tencent Cloud database TDSQL, through easy de-O, mixed load to meet multiple scenarios, through high performance, low unit price to help institutions reduce costs and increase efficiency, and through intelligent operation and maintenance to improve operation and maintenance efficiency.

In 2023, TDSQL broke the TPC-C world record with a performance data of 814 million transactions per minute, using only one-third of the industry's unit cost.

After years of development, TDSQL's market share has ranked first in the financial industry, moving from the technical verification stage to rapid growth.

Taking the Agricultural Bank of China as an example, Tencent Cloud TDSQL helped the Agricultural Bank successfully adopt the core system construction supported by the database's native distributed capabilities for the first time among state-owned large banks, achieving a super-large-scale deployment of two places and three centers, 1800 nodes, saving more than half of the server quantity.

The server operating system is the most important core system in the computer, which is the key to ensuring the safe, stable, and reliable operation of the business system.

TencentOS Server, which Tencent started developing in 2010, supports full-stack national cryptography and has smoothly supported its own businesses such as WeChat, games, and payments for more than 10 years with tens of millions of nodes.

TencentOS Server V3, which was released in August this year, supports on-site replacement of CentOS and has made many upgrades for cloud and AI scenarios, which can increase GPU utilization by 2 times.

Currently, TencentOS Server has become the largest deployed Linux operating system in China with a deployment scale of nearly 10 million sets and has been running stably on a large scale in dozens of banks, securities firms, and consumer finance companies.

Moreover, both TDSQL and TencentOS Server have passed the national security and reliability evaluation.

1.

Tencent Cloud TCS Cloud-Native Suite: Fully helping financial institutions transform to cloud-native.

Cloud-native is the inevitable trend of distributed upgrades, which can greatly improve resource utilization.

By microservice-based business system distributed upgrade and transformation, it can greatly improve R&D and online efficiency, with excellent scalability and observability.

However, the threshold for the implementation and implementation of cloud-native technology is relatively high, and there are the following four major difficulties: (1) Lack of mature methodologies such as route planning, financial investment, and supporting reforms; (2) Self-built and open-source are difficult to meet financial-level cloud-native requirements, and lack platforms and tools; (3) Microservices/unitization need standard guidance, lack of experience; transformation and migration need ISV support & adaptation; (4) Need clear evaluation points, continuous optimization, improvement, lack of methodologies and practical experience.

Tencent Cloud TCS Cloud-Native Suite, born from the accumulation of Tencent's self-developed business support, includes rich components such as microservice frameworks, distributed messages, distributed caches, and DevSecOps, supporting one-stop cloud-native service capabilities from design, development, testing, delivery to operation and maintenance on independent innovation infrastructure.

It has already supported Tencent to achieve full cloud-native of group business in three years, with a volume of more than 50 million cores, which is the largest scale of cloud-native practice in China, saving billions of costs a year.

Tencent Cloud combines years of its own practice to provide Tencent's plan for financial institutions to transform to cloud-native.

It provides full-process services in strategic planning, tool platform construction, technical specifications, and application transformation and migration, and helps institutions continuously improve and optimize based on Tencent's "Cloud-Native Maturity Model."

Currently, Tencent Cloud has helped many state-owned large banks, joint-stock banks, regional banks, leading insurance companies, and securities firms to put their core businesses into production and go online.

In the practice of a leading bank in Southwest China, it adopted Tencent's full-stack financial-level cloud-native solution, and through unitized architecture, it achieved full cloud-native of more than 100 key systems including core, channels, and payments; through cloud-native resource scheduling and management, resources saved more than 60%.

2.

Tencent Cloud's new generation of big data platforms: lighter, faster, easier to use, and safer.

With the advent of the intelligent era, data elements have become the "means of production" of digital finance.

How to fully activate the potential of data elements and release the value of data elements has become a focus of the financial industry.

Currently, a large number of financial institutions are actively investing in the construction of independent innovation of big data infrastructure.

On June 18th this year, Tencent Cloud's big data TBDS was newly upgraded, and through deep integration of multi-architecture technology, it achieved lake and warehouse unification on a single platform.

Tencent Cloud's new generation TBDS provides full-scenario data service capabilities for the entire data lifecycle, which is lighter, faster, easier to use, and safer, with financial-level high performance and high reliability.

After the upgrade, TBDS's comprehensive performance has increased by 20%, and storage and computing can be scaled and intelligently optimized on demand.

In the insurance industry, Tencent Cloud's big data TBDS has jointly implemented the first domestic data lake and warehouse integration practice in the insurance industry with Taiping Life, effectively improving the efficiency of Taiping Life's offline computing, real-time computing, and ad hoc data analysis scenarios.

The data output time has been accelerated from more than 1 hour to about 5 minutes, and the aggregation query scenario has been accelerated by 5 times.

Currently, Tencent Cloud's big data TBDS has been implemented by more than 2,000 customers.

3.

AI large model application: expanding the service boundaries of the financial industry.

In the past two years, China's AI large model industry has been booming.

The financial industry has a series of advantages such as abundant structured data, rich application scenarios, and good digital infrastructure, and is considered a good harbor for the industrial landing of AI large models.

In 2023, when Tencent took the first step in large models, it insisted on doing "AI closest to the industry," integrating AI into industrial scenarios.

Currently, the Hunyuan general large model has been implemented in more than 700 business scenarios of Tencent.In the financial industry, Tencent Cloud has trained a financial industry large model on top of the Hunyuan General Large Model, using a vast amount of financial vertical information, equipped with professional financial scenario task capabilities, currently covering front, middle, and back offices as well as comprehensive operations and other key areas of application.

(1) Tencent Cloud AI Code Assistant Tencent Cloud AI Code Assistant is based on the Hunyuan Large Model, combined with its own massive code library and open-source code, fine-tuned to create four major application scenarios: code completion, technical conversation, automated testing, and code diagnostics, which can greatly enhance development efficiency.

Currently, 70% of programmers at Tencent use the code assistant in their daily work.

In the financial industry, the code assistant is one of the most widely applied scenarios of AI large model implementation, with over 6,000 programmers using it in the largest financial institutions.

Tencent Cloud AI Code Assistant supports enterprise fine-tuning training and also supports rapid integration with enterprise code libraries through RAG (Retrieval-Augmented Generation).

Among the cooperating customers, a general character generation rate of 35% and a code adoption rate of over 30% can be achieved; at the same time, efficiency can be improved through knowledge Q&A and code diagnostics, testing capabilities, increasing overall coding efficiency by more than 43%.

(2) Knowledge Engine and Customer Service In the knowledge engine and customer service scenario, Tencent Cloud relies on the knowledge raw materials of financial institutions to build a knowledge engine base for customers, and then combines the conversational Q&A capabilities of the financial large model with digital humans to form a new generation of intelligent customer service products, proficient in customer service, outbound calls, brand promotion, Q&A, marketing, and other modes.

At the same time, through RAG (Retrieval-Augmented Generation), the accuracy and relevance of responses are greatly improved, effectively reducing illusions.

The intelligent customer service empowered by the large model can provide 7x24 service, solving more than 90% of complex problems.

The cost of conversation construction is reduced by two-thirds, significantly saving labor costs and operational difficulties.

It is currently widely used in top insurance and banking financial institutions.

(3) Financial Risk Control Large Model In addition, the risk control field is also an important scenario for the implementation of large models, with significant quality and efficiency improvements.

Tencent Cloud integrates the large model into the model construction process, based on a vast amount of risk control knowledge, through extensive pre-training and knowledge distillation, helping financial institutions to quickly build customized anti-fraud risk control models according to their own credit scenario characteristics and product customer group features.

The traditional risk control model construction cycle of two weeks is shortened to two days, and the online cycle is reduced from seven days to hours.

In practical implementation, Tencent Cloud has helped a bank customer build a risk control large model in the credit card loan scenario, with a daily query volume of up to 5 million times, and 50,000 fraud victims are identified daily, of which nearly 70% are scams involving impersonating shopping e-commerce logistics loan customer service.

The warning data accuracy rate is as high as 98%, and Tencent Cloud has helped the bank avoid fraudulent transactions amounting to hundreds of millions of dollars.

(4) AI Application Scenarios in the Securities Industry Finally, in the securities industry, Tencent Cloud has also explored some AI application scenarios.

With the capabilities of large models, RAG (Retrieval-Augmented Generation), and financial sentiment, Tencent has conducted in-depth exploration of scenarios within its internal financial technology system and on the customer side.

In investment research, it is possible to quickly interpret individual stock movements, increasing coverage by 50 times and speed of reach by 5 times compared to manual editing; in research reports, it can extract key points from financial reports in minutes and provide insights; in sentiment, it supports real-time sentiment Q&A, and automatically forms integrated sentiment reports; in investment consulting, it enhances the professional baseline of investment consulting and expands the service radius of expert investment consulting, and so on.

Basic software is a key infrastructure of the digital economy, and its development level is related to national and industrial security.

However, every breakthrough behind it requires countless efforts and difficult challenges.

Over the past few years, Tencent has built an IT infrastructure integration and innovation ecosystem centered on the computing industry, gradually covering key nodes in the ecosystem chain, from server operating systems, databases to big data platforms.

The acceleration of independent innovation product capabilities and the implementation of industrial scenarios are a key link.

Tencent Cloud hopes to work with ecological partners to deeply cultivate the field of integration and innovation.

On the one hand, it will comprehensively upgrade the distribution channel system and develop a number of professional partners with experience in sales and service of private products; on the other hand, Tencent Cloud will also establish a "Convergence Innovation Open Alliance" to carry out joint adaptation, joint expansion, and joint innovation with more partners, promote the integration of joint solutions and products, create benchmark cases of convergence innovation, and create value for convergence innovation in various industries.

Today, AI technology represented by large models has become a new quality productive force driving the high-quality development of the financial industry.

Tencent Cloud is also actively exploring more AI large model implementation scenarios in the financial industry, working with joint customers and ecological partners to embrace the "large model moment" of China's financial industry.

In the future, Tencent will continue to strengthen independent R&D investment, refine product and solution capabilities, help the transformation and upgrading of China's industry, and contribute scientific and technological strength to the development of China's digital economy.