On April 18, DingTalk officially launched the AI assistant market (AI Agent Store), and the first batch of more than 200 AI assistants will be launched, covering enterprise services, industry applications, efficiency tools, finance and taxation legal affairs, education and learning, life and entertainment and other categories.
Now, users can search for "AI assistant market" in DingTalk to choose to enable each AI assistant.
At the same time, DingTalk released a set of data: since April 18, 2023, it was announced that it would access the Tongyi Qianwen model and start a comprehensive intelligent strategy, DingTalk AI assistant, document AI, Yida AI, intelligent OA and other products are penetrating into thousands of industries. As of the end of March 2024, DingTalk AI has been used by more than 2.2 million enterprises, with more than 1.7 million monthly active enterprises.
4 categories of 200+ AI assistants are the first to be put on the AI assistant market, out of the box
What distinguishes DingTalk AI from large model companies is that it has rich enterprise scenarios, data accumulation, and ecological applications. The AI assistant market connects the SaaS applications and low-code application ecosystems accumulated by DingTalk, and connects enterprise-built systems and external third-party platforms through open capabilities, so that everyone and every enterprise can create AI assistants with a low threshold.
The first batch of 200+ AI assistants on the shelves of DingTalk can be divided into four categories in terms of the richness of their action capabilities:
- Role: AI assistant - everyone's life, entertainment, and office assistant. They can talk to large models and search and answer information in specific areas in real time, such as computer gurus who focus on solving various hardware and software problems, such as computer jams, network instability, etc., and travel assistants are dedicated travel planners who can plan routes, check the remaining number of tickets for high-speed rail, or information about global attractions, hotels and restaurants.
- Professional AI assistant - An AI assistant that completes professional knowledge or industry knowledge training on the basis of a large language model. For example, "Xiaozhu" has learned the professional knowledge of the construction industry, and users can query the details of the construction industry like chatting, and have no problem finding documents, reading information, and finding professional policies;
- Multi-tasking AI assistants, such as the public security government service assistant built by the Hangzhou Municipal Public Security Bureau, can not only answer government services, such as one-stop services from residence permit application to entry and exit business, but also open up a number of applications, which can add personal schedules or fill out opinion collection forms according to the needs of the dialogue.
- Cross-application AI assistants - More than 30 AI assistants provided by SaaS service providers such as Yonyou Payroll, Mojian, and Ctrip Business Travel not only have Q&A, professional knowledge and other capabilities, but also open up the original SaaS applications. For example, the "Salary Analysis Assistant" built by Yonyou Payroll can efficiently query labor costs and personnel flows, and provide market salary analysis, so that enterprises can enhance their competitiveness while controlling costs;
At present, the DingTalk AI assistant market adopts the mode of "application and review on the shelf", and high-quality AI assistants that meet the standards can be put on the market and shared with all users for selection.
The AI action system has been upgraded to enable anyone to create practical AI
"The ability to act is the core difference between DingTalk AI and simple content creation and Q&A AI. After being able to take action, the AI assistant can connect with various applications, provide scenario-based intelligent services, and truly enter scenarios such as application, collaboration, and operation. The assistant person in charge of DingTalk AI said.
With the launch of the market, DingTalk has also upgraded the action system of the AI assistant, which supports three development methods: anthropomorphic operation, workflow, and custom capabilities, making the connection between AI and applications simpler.
- "Anthropomorphic Operation" allows AI to learn and simulate a human to operate an application. The user demonstrates the AI once, and the AI assistant can learn the user's operation process and simulate the operation process without any code. In addition, anthropomorphic operations support generalization capabilities, so you can draw inferences from each other and deal with the same type of operations without having to learn again.
- "Workflow" aims to solve complex and multi-link tasks, and through the orchestration of AI execution processes, multi-link operations can be completed automatically and step by step, so that AI assistants can perform complex batch tasks. Workflows provide multiple types of components, including web page access requests, API interface calls, DingTalk functions such as scheduling and messaging, and the selection of all DingTalk connectors.
- To meet the high-level needs of professional developers, the "Customization Capability" also supports code development, complete the call of enterprise stock applications, or connect to various third-party platforms such as video, information, and e-commerce outside DingTalk. For example, the travel AI assistant created by the enterprise can complete the work of booking air tickets, hotels, and itinerary on third-party platforms according to instructions.
Tongyi Qianwen and Tongyi Farui entered the AI assistant market
In addition, two large model applications, Tongyi Qianwen and Tongyi Farui, have been put on the DingTalk AI assistant market. Tongyi Qianwen AI assistant can interact with text or voice, allowing the large model to provide multimodal services such as Wensheng text, Wensheng diagram, and image understanding.
Tongyi Farui is an AI assistant with understanding and reasoning capabilities in the legal field, which can conduct conversations with users based on natural language, accurately answer legal questions, push adjudication cases, assist in case analysis, generate legal documents, and intelligently retrieve laws and cases.
Al assistant is becoming the mainstream form of future applications, and will surely form a rich business and ecological model. The DingTalk AI assistant market is committed to becoming the most active AI assistant incubation, distribution and trading platform, and the official DingTalk Al assistant, the Al assistant built by ecological partners and developers, and the Al assistant created by individual users will become the three main components, and obtain commercial benefits from them, and everyone is a creator and entrepreneur of AI applications.