Tencent’s collaborative SaaS products are fully connected to the mixed model to realize intelligent upgrade.

Large-scale model application will be accelerated after landing.

On April 22nd, Tencent announced that its collaborative SaaS products were fully connected to Tencent’s mixed model. In addition to "three outstanding products" such as enterprise WeChat, Tencent Conference and Tencent Document, collaborative SaaS products such as Tencent Lexiang, Tencent Electronic Sign, Tencent Questionnaire and Tencent Cloud AI code assistant have also been intelligently upgraded.

Tencent Hybrid has highlighted its practical label since its launch, and one of its core missions is to optimize the use experience of Tencent products. At present, Tencent’s mixed-element model has been expanded to the scale of trillions of parameters, and it is the first in China to adopt the mixed expert model (MoE) structure. It is better at dealing with complex scenes and multi-task scenes, and its overall Chinese performance is at the leading level in the industry. Tencent’s hybrid element has excellent performance in mathematics, code, logical reasoning, multi-round dialogues and literary graphs. At the same time, it provides models of different sizes to adapt to more application scenarios that require low cost and high reasoning performance. External developers and enterprises can directly call Tencent’s meta-mixing ability through the API on Tencent Cloud. These capabilities combined with SaaS products can solve the pain points of users in specific scenarios.

In the enterprise internal learning and training scenario, Tencent enjoys combining the big model ability with the enterprise internal knowledge, and using AI assistants to assist learning to improve the efficiency of knowledge transfer; In the scenario of electronic contract, Tencent Electronic Sign improves the efficiency of contract drafting and review and intelligently screens risks with the help of big model and AI capabilities. In the questionnaire survey scenario, Tencent Questionnaire has created an AI solution for the whole process, covering the design, delivery, recycling and analysis of the questionnaire, greatly improving the quality and efficiency of the survey. In the development scenario, Tencent Cloud AI Code Assistant provides AIGC services for developers, development teams and corporate customers with safe, compliant and high-viscosity experience based on Tencent’s professional coding model.

In the collaborative office scenario, Tencent Conference will hand over miscellaneous meeting minutes, course notes, to-do lists and other matters to AI assistants and intelligent recording, so that participants can effectively improve their concentration and make communication more efficient; Tencent documents rely on the ability of large model to allow users to quickly process various documents and create efficiently through AI intelligent assistant with one sentence instruction.

Tencent enjoys: AI assistant assists learning and improves the efficiency of knowledge transfer.

Tencent Lexiang is a knowledge learning collaboration platform serving enterprises, providing learning and training, knowledge management and cultural construction services for enterprises. At present, it has served more than 300,000 customers and authorized more than 100 million users.

Based on Tencent’s mixed model and knowledge engine, Lexiang launched an AI assistant to help enterprises improve the efficiency of knowledge production and learning. Lexiang AI Assistant integrates the big model ability with the internal knowledge of the enterprise to provide intelligent question and answer, intelligent generation and auxiliary creative ability.

Compared with the traditional search mode, AI intelligent question answering can integrate multi-modal information such as product documents and videos, summarize and generate natural language answers, and further shorten the information acquisition link.

In the training scenario, the AI intelligent generation function can summarize the knowledge points of generating live content and courses to help employees improve their learning efficiency. At the same time, it can automatically generate examination questions and tests according to the training content, helping the trainers to test the training effect.

The auxiliary creative function can continue to write, polish, abbreviate, correct mistakes, etc. according to the content of the document to help users generate materials quickly.

In industries with high knowledge threshold, Lexiang AI assistant significantly improves the efficiency of knowledge transfer. The Kuangzhen Law Firm in Hunan imported more than 40,000 internal knowledge documents compiled by employees into Tencent Lexiang, built an exclusive AI knowledge base, and realized the AI intelligent question-and-answer function. Employee surveys show that the satisfaction of AI answers to typical questions is as high as 93 points, and the accuracy of end-to-end questions is 91%.

Tencent Electronic Sign: Improve the efficiency of contract drafting and review, and intelligently screen risks.

Tencent e-sign is an electronic contract management tool created by Tencent. Because it can quickly initiate and sign contracts based on the WeChat ecosystem, it has grown rapidly since its launch, and has served more than 30 million individuals and corporate users.

After accessing Tencent’s mixed model, Tencent Electronic Sign provides intelligent support in the whole life cycle of contract drafting, review, signing and management.

When drafting a contract, users can input the contract requirements in natural language and directly generate the contract through Tencent. In the review process, the ability of text summarization is used to refine the approval opinions of different roles, which is convenient for participants to quickly review the contract. At the same time, combined with PDF parsing and document intelligent multimodal ability, the contract content can be presented in a structured way, and the differences between multiple contracts and versions can be accurately compared, thus improving the efficiency of contract review. At the risk screening level, the AI intelligent document review system is used to accurately identify contract risk clauses and generate the results of AI review, which is convenient for enterprises to control contract risks.

Enterprises have different requirements for risk control of contracts. Tencent e-sign also uses large model and few-shot technology to train small vertical models suitable for customer industries to achieve low-cost operation. At the same time, through the hybrid cloud model, it supports the privatization deployment of data and models, solves the efficiency problem and ensures compliance.

In government departments with many contracts and documents, the intelligent ability of electronic signature has significantly improved the efficiency of contract management. A city government has built a "smart contract" system based on Tencent’s electronic sign, and introduced intelligent comparison and risk screening capabilities in contract and agreement scenarios such as party organization moving in, building approval, land lease, and private charging pile installation application. The test data shows that its approval efficiency has increased by 97%.

Tencent questionnaire: intelligent design, delivery and analysis, greatly improving the efficiency of research.

Tencent Questionnaire is a professional online questionnaire survey platform under Tencent, which provides professional investigation and research solutions based on data collection and has over one million real sample users. Up to now, the number of users of Tencent’s questionnaire service has exceeded 48 million, and the number of registered teams has exceeded 200,000, helping users to recycle 3.4 billion questionnaires.

Based on the big model, Tencent questionnaire has created an AI solution for the whole process, covering the design, delivery, recycling and analysis of the questionnaire, greatly improving the quality and efficiency of the survey.

In the questionnaire design, users can directly describe their needs in natural language. Tencent’s AI solution can automatically generate customized questionnaires, including questionnaire survey, information reporting, online examination, 360-degree evaluation, voting, registration punching, online evaluation and many other types.

In the questionnaire delivery process, AI can realize the automatic delineation, delivery and screening of samples. In the process of launching, the large model can also generate targeted questions according to the needs of the researchers, screen the samples, and eliminate the samples that do not meet the requirements, thus improving the accuracy of launching.

After the questionnaire is completed, AI can identify invalid questionnaires and eliminate them, and generate a comprehensive and professional analysis report with one click, covering frequency analysis, text analysis and other conclusions, effectively improving the analysis efficiency. At the same time, AI can also cluster and summarize the subjective text questions in the questionnaire to help the questionnaire issuer quickly understand the respondents’ views and concerns.

AI Code Assistant: Strengthen coding AIGC service to help developers improve quality and efficiency.

Tencent Cloud AI Code Assistant is a digital tool for intelligent completion and generation of positioning codes. Based on Tencent’s professional coding model, it can provide AIGC services for developers, development teams and corporate customers with safe, compliant and high-viscosity experience. It has the functions of code completion, technical dialogue, code diagnosis and unit testing.

In terms of code completion, AI code assistant can provide automatic completion with good experience based on multi-file understanding and context, effectively improving coding efficiency and quality. In terms of technical dialogue, it can help developers to ask and answer technical questions at any time and give technical answers with reasoning value; It also supports inline dialogue and code interpretation, guiding developers to accurately understand the engineering context and guiding problem solving. In the aspect of code diagnosis, it can help developers diagnose code problems, give suggestions for correction, and at the same time, find problems that rules can’t find, and provide valuable and readable suggestions. In unit testing, unit test cases can be generated and executed according to the executable test code, which can reduce the time of writing single test and improve the quality and efficiency of developers’ development and learning work as much as possible.

It is understood that Tencent Cloud AI code assistant has achieved more than 40% employee coverage in a cooperative enterprise, with a code generation rate of about 30%, an adoption rate of about 20%, a warehousing rate of over 60%, and a completion time of less than 500 ms.. Combined with the internal large-scale production experience of Tencent Cloud, the coding efficiency has been improved by more than 20%.

Enterprise WeChat: It has been put into use in email and customer service scenarios.

As an enterprise communication and office tool created by the WeChat team for enterprises, enterprise WeChat has connected over 12 million real enterprises and organizations, and the number of WeChat users through enterprise WeChat service reaches 500 million every day. As more and more enterprises build private domain business positions based on enterprise WeChat, the scale of private domain transactions through enterprise WeChat has reached hundreds of billions of yuan.

At present, enterprise WeChat has been connected to Tencent’s mixed model to enhance the product experience. In the enterprise mailbox, Tencent mixed-element model has been used to check spelling and grammar when English is input, which improves the accuracy; In the customer service scenario, enterprise WeChat combined with Tencent’s mixed model provides better service for customers and improves service efficiency.

Tencent conference: AI assistant, intelligent recording and real-time subtitle translation make the conference discussion more immersive.

In the past, users were always busy taking notes and writing notes during meetings or classes, which often distracted their attention from the contents of meetings or courses, resulting in low efficiency.

Tencent Conference continues to invest in intelligent technology research and development and capacity-building, handing over miscellaneous meeting minutes, course notes, to-do items and other matters to AI assistants and intelligent recording, so that participants can effectively improve their concentration during the conference and devote themselves to the conference content itself, so that they can think about problems more actively, understand content and give feedback, and make communication more efficient.

This time, Tencent Conference demonstrated more AI technical capabilities, such as realizing personalized reminders and refining key information through AI assistants to help users easily cope with complex and huge course and conference information; Check the contents of intelligent transcription and full-text summary in time after the meeting through intelligent recording, understand the meeting minutes and backlog, and help users quickly find the target topics and contents through intelligent chapters or speakers. In addition, real-time subtitle translation can also reduce the user’s understanding cost when facing cross-language conferences or courses.

Tencent document: "One sentence" opens a new experience of efficient creation

As a cloud efficiency tool that can be edited collaboratively, Tencent Docs currently has more than 200 million monthly active users, and the cumulative number of documents created by users exceeds 5 billion, and the number of service enterprises and organizations exceeds 1.2 million.

Relying on the AI model, Tencent Document Intelligent Assistant has been fully applied to documents, tables, slides, PDF, intelligent documents, collection tables, mind maps and other document types, covering many capabilities such as second-level processing of text content, application of function formula operation, accurate presentation of table data, rapid beautification of PPT generation, automatic analysis of collection results, one-click generation of mind maps and so on, and supporting the smooth flow of cross-category document content. Users can open the intelligent office and efficient creative experience after AI fusion through the instruction of "one sentence" on the computer side or mobile side of Tencent Document.

At present, more than 400 businesses and scenarios within Tencent have been connected to Tencent’s mixed model, and remarkable progress has been made in product innovation and cost reduction and efficiency improvement. Tencent Hunyuan is also actively developing multi-modal models, further strengthening the capabilities of Wensheng maps and Wensheng videos, providing services for more scenes and products, and helping enterprises to achieve intelligence.