Whether it's AI Agents in Web2 or Web3, they are all built on large language models (LLMs) as the core technological foundation. It can be said that the rise of LLMs has made AI Agents a reality and paved the way for their commercialization. After early technological accumulation, LLMs experienced an application explosion in 2022. OpenAI released GPT-3.5, which made significant advancements in natural language processing technology, but at this time, AGI had not yet reached the level of an "Agent." Starting in 2023, many companies around the world began to release multiple open-source large language models (LLMs), including LLaMA, BLOOM, StableLM, and ChatGLM. AI Agent creation platforms like Voiceflow also emerged, and by August 2023, over 130,000 teams were building AI Agents on Voiceflow. After 2024, competition among LLMs gradually intensified, with upgraded versions and new LLMs, including Gemini 2.0 and DeepSeek R1, being continuously launched. The gradually decreasing usage costs, increasingly powerful analytical capabilities, and more AI Agent tools brought about by competition have encouraged more people to get involved. As the performance of LLMs continues to improve, AI Agents are no longer just a concept in the research field but have become practical tools in reality. In 2024, Microsoft integrated 10 autonomous AI agents into Dynamics 365, which can automatically complete tasks in customer service, sales, finance, warehousing, and other processes. At the end of the same year, Google released the multimodal large model Gemini 2.0 and launched three new agent prototypes based on it, including the general large model assistant Project Astra and the programming assistant Jules. In 2025, OpenAI launched the first AI agent Operator, capable of automatically completing complex operations such as writing code, booking travel, and e-commerce shopping. As LLM companies directly launch AI Agents, their core purpose is to attract developers and builders to use their LLM technology, expand application scenarios, and increase revenue. This transition from research to application also marks a new development stage for AI Agent technology.