Xueersi launches Precise Learning Lobster "Little Dragon"

On April 1, Xueersi released the industry’s first native student-side precision learning lobster product, custom-built based on the OpenClaw architecture—Xiao Jinglong. With “brain mapping, private property, and learning companion” as its core positioning, Xiao Jinglong for the first time integrates long-term memory, dynamic learning-situation diagnosis, an education-dedicated Skill chain, and an emotional companionship system into a student-dedicated intelligent agent, providing students with end-to-end, all-subject AI study companionship services. The product is currently available for download via the official website https：//mate.tal.com/.

In 2026, as the open-source intelligent agent OpenClaw quickly goes viral worldwide, Agentic AI is moving from “being able to answer” to a new stage of “being able to perceive, plan, execute, and reflect.” Its capabilities naturally align with modern education scenarios underpinned by learning-situation insights, intelligent interaction, and individualized instruction, making education AI go beyond passive Q&A and evolve toward “a student-dedicated intelligent agent for long-term growth.”

Seamlessly integrating AI into continuous learning scenarios, forming a dynamic mapping of the learning world

According to the introduction, “Xiao Jinglong,” built on Xueersi’s 22 years of educational experience and industry-leading education AI capabilities, connects the key links across a student’s entire learning lifecycle, and is equipped with an “education-dedicated Skill chain.” It can dynamically generate personalized learning and practice plans based on the learning situation, intelligently generate customized practice content, grade problems and provide targeted explanations around weak points, automatically accumulate customized summaries and a personal learning record, and build an immersive learning-companion scenario for students. Meanwhile, it also has abilities such as daily reminders, progress management, emotional perception, and interactive companionship—enabling AI to truly enter a student’s everyday continuous learning scenario for the first time. It can uncover knowledge gaps, capability weaknesses, cognitive sticking points, and the next optimal learning action behind every student action.

Students can talk with their AI learning companion, Xiao Jinglong, at any time. Whether it’s explaining problems, diagnosing learning situations, grading papers, or generating intelligent questions—so long as the student states their needs in the chat box, the system can automatically call the relevant Skill from the Skill marketplace. Interactive notes will organize and push the corresponding knowledge points, textbook chapters, demonstration videos, and more in real time. It’s like discussing problems with a real learning buddy: you chat and interact lightly, while also fully internalizing the knowledge.

The knowledge map of “Xiao Jinglong,” mapping the student’s situation

Inside “Xiao Jinglong,” various high-level learning abilities are decoupled and packaged into independent Skill plugins. Under this architecture, students do not need to repeatedly switch functional entry points or do complex configuration. Xiao Jinglong can independently perceive context, dynamically orchestrate Skills, and automatically provide the corresponding capabilities based on the student’s needs. A related负责人 explains that, through the education-dedicated guardrails and a “clean intelligent agent” design, it can effectively control large-model hallucinations and interference from irrelevant information. On the other hand, by re-structuring the “education-dedicated Skill chain,” the intelligent agent gains “hands and feet” capabilities—dynamic calling abilities oriented to different learning tasks. What Xiao Jinglong presents is not “multiple functions placed side by side,” but an education intelligent agent system that works autonomously around the student’s task goals.

Long-term memory capabilities empowered, preserving “cognitive assets” for growth

Long-term memory capability is also one of Xiao Jinglong’s most core differentiated barriers. The foundation of precise learning lies in comprehensive, detailed understanding of students’ learning situations. If learning data remains at a static recording level, it becomes difficult to truly drive precise learning.

Xiao Jinglong has established a short- and long-term memory system for student growth. It can continuously write multi-modal cues—such as dialogue feedback, Skill call results, practice performance, hesitation pauses—into each student’s dedicated memory stream. As a result, it is no longer a “no-memory encyclopedia,” but more like a digital brain that gradually forms a student’s cognitive profile. It remembers the questions students asked earlier, can also anticipate potential future difficulties, and thereby generate more precise, more efficient, more individualized learning paths. Interactions between students and AI are no longer just an immediate consumption of effort, but an ongoing foundation infrastructure for growth—constantly accumulating and steadily appreciating.

“Xiao Jinglong’s” exam paper analysis capabilities

At the same time, Xiao Jinglong also introduces an emotion computation mechanism optimized with educational psychology into the Agent architecture. When the system identifies that a student is experiencing states such as setbacks, anxiety, or losing patience, it does not simply present standard answers. Instead, it uses more gradual guidance, pace adjustment, and encouragement feedback aligned with learning psychology principles. It adjusts the style of interaction and companionship based on different students’ age stages, personality traits, expression habits, and learning states. Every student can get a “learning companion” that understands them best.

Putting people first, returning to people—AI becoming the intelligent link between education and care

In March, Xueersi was the first to release “Nine-Chapter Dragon Lobster,” designed specifically for the teacher community. The unveiling of “Xiao Jinglong,” which is aimed at students, is precisely another step forward in Xueersi’s “Dragon Lobster Matrix” layout in the AI education field.

From empowering teachers to accompanying students, the “Dragon Lobster Matrix” is committed to connecting every part of education, so that AI is not just a single efficiency tool, but an intelligent link that ties together schools, students, families, and all parties across different scenarios. With continuous product planning, Xueersi is working to make “mass individualized instruction tailored to each student” become real through repeated technology deployments.

As answering, grading, and generation gradually become industry baseline capabilities, the value points of next-generation education products will return to “people” themselves—understanding students, consolidating students’ learning, and accompanying students. The future form of Education + AI should be a continuously growing, accumulating, and evolving intelligent system. When education AI moves from “being able to answer” to “being able to understand, remember, and accompany,” the era of student-dedicated intelligent agents truly begins.