Baidu's Lingjing Matrix Upgraded to Intelligent Agent Platform, Ushering in an Era Where Everyone Can Develop Agents
-
AI Home News, December 18th: Baidu's 'Lingjing Matrix' platform has been upgraded to the 'ERNIE Large Model Intelligent Agent Platform'. Based on the ERNIE large model, Lingjing Matrix offers developers diverse development approaches, enabling them to choose methods tailored to their industry domains and application scenarios to create native applications for the large model era.
Lingjing Matrix also boasts the most comprehensive intelligent agent ecosystem in China. Backed by the powerful ERNIE large model, it has already attracted over 30,000 developer applications. Leveraging Baidu's full-scenario ecosystem, developers can access more traffic distribution channels and business opportunities. Currently, numerous agents such as legal assistants, TreeMind mind maps, and AI-powered resume tools like Workplace Password have successfully navigated the path from development to distribution and monetization through Lingjing Matrix.
From an industry-wide perspective, the battle of large models has gradually shifted from the model layer to the ecosystem and application layers. At the GeekPark Innovation Conference 2024 held on December 16th, Robin Li, Founder, Chairman, and CEO of Baidu, stated that the value of large models can only be realized when there are countless AI-native applications built upon the foundational models.
Among various applications, intelligent agents (Agents) are considered one of the most promising categories. It is widely recognized that agents built on large model capabilities possess 'independent thinking' abilities, enabling them to utilize various tools to progressively achieve their objectives.
Regarding the 'upgrade' of Lingjing Matrix's intelligent agents, Zhang Ruixing, head of the Lingjing Matrix platform, said, 'Lingjing Matrix initially focused on serving plugins, but as the demand for intelligent agents in the industry grew, we upgraded accordingly. Whether for plugins or agents, Lingjing Matrix aims to help developers reduce development costs while providing traffic distribution channels to complete the commercial loop.'
In terms of development, Lingjing Matrix, as an intelligent agent platform, significantly lowers the barrier to agent creation. Here, developers can use natural language interaction to develop agents quickly and efficiently, with some agents created in just minutes. Currently, Lingjing Matrix offers three modes—no-code, low-code, and full-code—to meet the needs of developers at different skill levels.
In addition to lowering development costs, Lingjing Matrix possesses China's most comprehensive intelligent agent ecosystem, empowering developers to succeed.
First, Lingjing Matrix leverages the country's most advanced Wenxin Large Model 4.0, providing developers with the optimal foundation for intelligent agent implementation. Second, shortly after its launch in September this year, over 30,000 developers applied to join, creating a rich ecosystem of large model plugins that help intelligent agents expand their capabilities and boundaries. Notably, Lingjing Matrix utilizes Baidu's full-scenario resources to offer developers corresponding traffic distribution channels and commercial opportunities.
Building on these features, Lingjing Matrix currently hosts over 2,000 live intelligent agents, enabling multiple partners to successfully navigate the complete path from low-cost development to distribution and monetization. In other words, the Lingjing Matrix platform not only creates new opportunities and scenarios for developers but also reduces customer acquisition costs, laying the foundation for subsequent monetization and private domain marketing by accumulating seed users.
For example, the "Legal AI Assistant" in the Lingjing Matrix Experience Center is a legal Q&A intelligent agent jointly developed by Baidu Search and the China Judicial Big Data Research Institute. It combines authoritative legal data with Baidu's large model to address users' daily legal queries. In implementation, it uses Lingjing Matrix's data access mode to integrate with the Wenxin large model at low development costs and distributes across various suitable scenarios. Currently, this legal Q&A agent receives nearly 43,000 daily visits, with cumulative visits exceeding 2.3 million since its launch three months ago.
Additionally, "TreeMind," an AI-powered mind mapping tool, is one of Lingjing Matrix's first partners and a successful case of achieving commercial closed-loop. After completing development and obtaining distribution through Lingjing Matrix, TreeMind now handles over 50,000 daily calls, with a user registration rate exceeding 60% from Baidu's traffic. The conversion rate for registered paying users reaches 3%, surpassing the industry average for tool monetization. Currently, over 10% of TreeMind's daily active users come from Baidu's distribution channels.
Similarly benefiting from low-cost development on the Lingjing Matrix platform, along with traffic distribution and reduced customer acquisition costs, are tools like 'Career Password AI Smart Resume' and 'QuickRead'. Currently, 30% of the new users for 'Career Password AI Smart Resume' come from Baidu, with a 20% increase in registration rates due to precise targeting, leading to a 30% growth in membership. The AI tool 'QuickRead', which parses content via URLs, has gained distribution from Baidu's ERNIE Bot Web interface and Baidu search scenarios after integrating with Lingjing Matrix, now handling over 30,000 daily calls. Overall, the types of agents on the Lingjing Matrix platform already cover various fields including office work, lifestyle services, and daily consumption. The number of agents on the platform continues to grow. Soon, more agents like 'Product Purchase Recommendations' for efficient decision-making and 'Pen Spirit' for long-form learning and office tasks will be launched, completing the closed loop from low-cost development to traffic distribution.