What capabilities will the intelligent upgrade of SaaS products improve for industries and enterprises?
The application of large models is accelerating.
On April 22, at the Tencent Cloud Collaborative SaaS Intelligent Upgrade Media Communication Conference, Tencent announced that its collaborative SaaS products are fully connected to Tencent's hybrid model, in addition to the "one door and three masters" products such as WeCom, Tencent Meeting, and Tencent Docs, Tencent Lexiang, Tencent Cloud AI Code Assistant and other collaborative SaaS products have also been intelligently upgraded.
On April 22, at the Tencent Cloud Collaborative SaaS Intelligent Upgrade Media Conference, Tencent announced that its collaborative SaaS products are fully connected to Tencent's hybrid model.
Since its launch, Tencent Hybrid has highlighted its practical label, and one of its core missions is to optimize the user experience of Tencent's products. At the event, Zhang Feng, head of Tencent Hybrid Model Application, said that Tencent's hybrid model has been expanded to a trillion-level parameter scale, taking the lead in adopting a hybrid expert model (MoE) structure in China, and is better at handling complex scenes and multi-task scenarios, and the overall performance of Chinese is at the leading level in the industry. Tencent MixElement provides excellent performance in mathematics, code, logical reasoning, multi-round dialogue, and Wensheng diagrams, and provides models of different sizes to adapt to more application scenarios that require low cost and high inference performance.
It is reported that at present, external developers and enterprises can directly call Tencent's hybrid capabilities through Tencent Cloud's APIs, so what pain points can these capabilities be solved for users after combining these capabilities with SaaS products?
WeCom:
Improve work efficiency for tens of millions of users
As a corporate communication and office tool created by the WeChat team, WeCom has connected more than 12 million real enterprises and organizations, and the number of WeChat users served by enterprises through WeCom has reached 500 million every day.
At present, WeCom has been connected to Tencent's hybrid model to improve product experience. In the enterprise mailbox, Tencent Mixed Yuan model has been used to check the spelling and grammar of English input to improve accuracy, and in customer service scenarios, WeCom combined with Tencent Mixed Yuan model to provide customers with better services and improve service efficiency.
Tencent Meeting:
Make meeting discussions more immersive
In the past, users were always busy taking minutes and writing notes during meetings or classes, which often distracted them from the content of the meeting or course, resulting in low efficiency.
Tencent Meeting continues to invest in the research and development of intelligent technologies and capabilities, handing over redundant meeting minutes, course notes, to-dos and other matters to AI assistants and intelligent recording, so that participants can effectively improve their concentration during the meeting, devote themselves to the meeting content, and think more actively about problems, understand the content and give feedback, and make communication more efficient.
This time, Tencent Meeting showcased more AI technology capabilities, such as personalized reminders and key information extraction through AI assistants to help users easily cope with complex and large course and meeting information, intelligent recording to view intelligent transcription and full-text summary content in a timely manner after the meeting, understand meeting minutes and to-dos, and help users quickly find target topics and content through intelligent chapters or speakers.
Tencent Docs:
"One sentence" opens up a new experience of efficient creation
As a cloud-based productivity tool that can be collaboratively edited, Tencent Docs currently has more than 200 million monthly active users, with more than 5 billion documents created by users, and more than 1.2 million enterprises and organizations served.
Relying on the AI model, Tencent Docs Intelligent Assistant has been fully applied to document types such as documents, tables, slides, PDFs, intelligent documents, collection forms, mind maps, etc., covering the second-level processing of text content, the application of function formula operations, the accurate presentation of table data, the rapid generation and beautification of PPT, the automatic analysis of collection results, and the one-click generation of mind maps, and supports the smooth flow of cross-category document content. On the desktop or mobile terminal of Tencent Docs, users can start the intelligent office and efficient creation experience after AI integration through the "one-sentence" command.
At present, more than 400 businesses and scenarios within Tencent have been connected to Tencent's hybrid model.
Tencent Lexiang:
Make knowledge transfer easier
Tencent Lexiang is a knowledge learning and collaboration platform serving enterprises, providing enterprises with learning and training, knowledge management and culture building services, and has served more than 300,000 customers and more than 100 million authorized users.
Based on Tencent's hybrid model and knowledge engine, Lexiang has launched an AI assistant to help enterprises improve the efficiency of knowledge production and learning. Enjoy the AI assistant integrates large model capabilities and internal knowledge of the enterprise to provide intelligent Q&A, intelligent generation, and assisted creation capabilities.
Compared with the traditional search mode, AI intelligent Q&A can integrate multi-modal information such as product documents and videos, summarize and generate natural language answers, and further shorten the information acquisition chain.
In training scenarios, the AI intelligent generation function can summarize and generate knowledge points of live content and courses to help employees improve learning efficiency. At the same time, it can automatically generate test questions according to the training content to help the trainer check the training effect.
The auxiliary authoring function can be continued, polished, abbreviated, and corrected according to the content of the document to help users quickly generate materials.
In industries with a high knowledge threshold, Lexiang AI assistant has significantly improved the efficiency of knowledge transfer. Kuangzhen Law Firm in Hunan Province has imported more than 40,000 internal knowledge documents compiled by employees into Tencent Lexiang to build an exclusive AI knowledge base and realize the AI intelligent Q&A function. According to the employee survey, the AI answers to typical questions are as high as 93 points, and the end-to-end question accuracy rate is 91%.
AI Code Assistant:
Strengthen the coding AIGC service to help developers improve quality and efficiency
Based on Tencent's professional coding model, Tencent Cloud AI Code Assistant is a digital tool that provides developers, development teams, and enterprise customers with AIGC services for safe, compliant, and high-viscosity coding scenarios, with functions such as code completion, technical dialogue, code diagnosis, and unit testing.
In terms of code completion, AI code assistant can provide automatic completion based on multi-file understanding and context, effectively improving coding efficiency and quality. In terms of technical dialogue, it can help developers to answer technical questions and answer questions at any time and give technical answers with inferential value, and also supports inline dialogue and code interpretation to guide developers to accurately understand the engineering context and guide them to solve problems. In terms of code diagnosis, it can help developers diagnose code problems, give suggestions for corrections, find problems that cannot be found by rules, and provide valuable and readable suggestions. In terms of unit testing, unit test cases can be generated and executed according to the executable test code, reducing the time for writing single tests, and improving the quality and efficiency of developers' development and learning work as much as possible.
It is understood that Tencent Cloud's AI code assistant has achieved more than 40% coverage of development employees within a cooperative enterprise, with a code generation rate of about 30%, an adoption rate of about 20%, a storage rate of more than 60%, and a completion time of less than 500ms. Combined with Tencent Cloud's internal experience in large-scale production, the coding efficiency has been improved by more than 20%.
At present, more than 400 businesses and scenarios within Tencent have been connected to Tencent's hybrid model, and significant progress has been made in product innovation, cost reduction and efficiency improvement. Tencent Hybrid is also actively developing multimodal models, further strengthening Wensheng graphics and Wensheng video capabilities, providing services for more scenarios and products, and helping enterprises achieve intelligence.
Written by: Superb
Editor and proofreader: Hu Yuan
Guidance: Xin Wen
Further reading: Jumping out of the bottomless pit of brute force computing: Wei Leping talks about the development and challenges of large models
Looking Back: Sora's Ten Questions for the New Year's Feature | 2024 Communication and AI Outlook: Butterfly Transformation in EmpowermentAI App Store: Will It Be "Fire"?Who is the AI Enabler: The 2023 AI Large Model Pioneer List TOP50 ReleasedWho is Emerging in Intelligence: The 2023 Industrial Large Model Pioneer List TOP20 is released
Peeling back the cocoon: What is the code capability of large models?
Communication Thinking I Meng Guozhu: What is the code ability of large language models?
Sixiang I Antelope Wen Weijun: The industrial model is the "core engine" to promote new industrialization
Compute for large models: NVIDIA unveils its latest GPUs
Here it is: the terminal platform that supports 10 billion parameter large models has been released
A "far ahead" open class: Robin Li teaches you "hand-in-hand" AI native applications
Volume large model, or volume application? Robin Li told you
Baidu Intelligent Cloud released Qianfan Large Model Platform 2.0: preset 41 industry datasets
The Open Tournament is coming: eight large models have been approved for launch
Expert Feature: Large Model Accelerates Full-Stack Terminal Subversion
The "100 model war" is in full swing: how to get out of the circle of operators' large models?
Large model meets terminal: what sparks are created?
Blockbuster Feature: Large Models Redefine Terminal Innovation
Three experts talk about the operator model: advantages, challenges and paths
Fierce Battle Model: The "Spear" and "Shield" of Operators
Sword pointing to industrial applications: Jingdong model is coming
TeleChat: The first large language model for operators is here!
Expert: Large models should not be filled indiscriminately, and general AI needs to be down-to-earth
Large-scale models: "universal is king" or "exclusive is gold"?
Large models open a new era of AI
A picture to see the large model
The confusion of AI for good: should you install a "traffic light" on the large model?
Who will lead the Chinese model?
Four experts talk: how can the communications industry focus on ChatGPT?
A picture to see the progress of "Chinese-style" ChatGPT
2024 Top 10 Technology Trends in the Communications Industry ReleaseWho is Leading the Way?2023 Top 10 Information and Communication Business Innovation Releases Tribute to the Frontline: 2023 Communications Industry Technology Innovation Figures Reveal Why Them?10 Annual Independent Innovation Product Technology Solutions
Hundreds of battles: The top 50 AI competitiveness list in 2023 is released, who is leading the way, and the top 10 cloud computer innovation is released