October 23, 2025 – Tencent’s IMA Open Day was held in Beijing, where the IMA 2.0 version was announced, scheduled to start internal testing on the 24th. The IMA 2.0 version includes features such as an Agent capability-based “Mission Mode” and the “AI Key Points” function for the knowledge base. The upgraded IMA will become a “collaborative partner” capable of understanding goals, executing tasks, and producing results.
Mission Mode: Auto-Generate Reports & Podcasts
The new “Mission Mode” in IMA supports generating two types of content: reports and podcasts. Users can initiate tasks using natural language queries on the homepage or within the knowledge base, and can also attach resources such as knowledge bases, documents, images, audio files, web pages, and notes to tasks, equipping the large model with “reference materials” while executing tasks. When generating audio-based podcasts, it supports selecting the number of participants in the conversation, voice tones, etc., meeting the needs for deep information processing and creation in learning or work scenarios.
After users activate “Mission Mode,” IMA will autonomously decompose and plan task steps using LLMs, based on web-wide or knowledge base data. It will call upon tools like content deep reading, web-wide search, knowledge base query, and content creation to complete user-input instructions. It self-monitors and corrects during execution, ultimately producing ready-to-use results. For example, inputs like “Help me write a new energy vehicle market report,” “Summarize the key points of the meeting discussion and generate minutes,” or “Design a teaching podcast on ‘Generative AI‘ for me” will be deeply learned and understood by the “Mission Mode” with Agent capabilities. It identifies the “hidden” tasks within the natural language, breaks them down into individual work nodes, and deeply utilizes the “comprehensiveness” of web-wide information and the “precision” of knowledge base materials, integrating the search, read, and write workflow.
Simultaneously, the knowledge base capabilities have also been upgraded. The new version introduces the “AI Key Points” function, which can automatically generate structured summaries. Multiple thematic conversations or tasks can be conducted in parallel within the same knowledge base, and it supports collaboration and sharing. New explicit indicators like “Likes” help users quickly assess the authority and activity level of a knowledge base.
200M Files Make IMA Your “Second Brain”
The latest news reported that in the year since its launch, IMA has deeply penetrated 20+ industries including technology, finance, education, healthcare, law, and government affairs. The top three fields by knowledge base quantity are: technology/internet, economy/finance, and health/healthcare. IMA has also gained deep adoption by real users, validating its practical value and broad industry adaptability, becoming a personalized “second brain” across various sectors.
In the medical field, Chief Physician Sun Xiaoli in critical care medicine has built IMA into a “medical second brain,” constructing a knowledge base based on the latest guidelines to quickly query medical knowledge, assist in formulating treatment plans, and achieve instant resolution of clinical problems.
In government services, Qianjiang Customs built an AI knowledge base using IMA, covering over 200 regulations and thousands of consultation examples, achieving nearly 100% accuracy in simulated Q&A, making policy responses both authoritative and actionable. Zhuhai Jinwan created the “Smart Jinwan Policy Assistant,” achieving real-time policy synchronization, plain-language interpretation, and precise matching, becoming a 24/7 online intelligent consultant for enterprises.
In educational innovation, a history teacher with 25 years of experience, Brother Ding, uploaded 30,000 teaching materials into IMA, turning the AI into an “exam point analyst” and “Q&A teaching assistant,” improving lesson preparation efficiency and student grades. The number of students ranking in the top 20 in his grade increased from 2-3 to 8. Professor Lu Jiangyong from the Guanghua School of Management at Peking University used IMA to build a course knowledge base, forming a “preview-interaction-expansion” teaching closed loop.
In finance, law, and scientific research, GF Securities created an “Investment Treasure Chest,” making professional knowledge widely accessible. Lawyer Wang Jiewen uploaded 778 legal precedents to IMA to extract judicial tendencies, reducing a task that originally took one week to just half a day. Qian Wenying, Secretary-General of the CEIBS AI and Management Innovation Research Center, uses IMA as a personal and team “second brain,” systematically integrating research results and building new infrastructure for scholar thinking and dissemination.
This is also verified in IMA’s development data. Over the past year, the total number of files in IMA knowledge bases has reached 200 million. Compared to January this year, monthly active users in September grew by over 80 times. IMA Product Lead Jayden stated: “We created IMA not just to solve individual efficiency problems, but hoping to build a secure levee amidst this flood of information, truly becoming your ‘Information Management Assistant,’ ensuring your information isn’t just ‘stored,’ but can be ‘grasped’ and ‘used effectively’.”
