On April 29, Tencent's intelligent workbench ima officially launched its personal knowledge Agent—copilot. This product aims to upgrade the existing AI search tool into a personal knowledge partner with continuous evolution capabilities through deep personalized memory and full-scenario perception. It marks a key transformation of AI applications from single-session conversation tools to long-term intelligent agents.

QQ20260429-112700.jpg

On the technical side, copilot integrates a self-evolving memory system composed of four core modules: "Soul (Persona), User (User Profile), Memory (Long-term Memory), and Agent (Experience and Skills)." It can structureively store user background and preferences across scenarios, eliminating redundant interactions caused by repeated input.

On the application side, copilot achieves full-scenario perception, functioning as a floating window within apps, directly understanding and processing current web pages, files, or notes in real time. In addition, the newly launched Skills ecosystem includes official skills such as cross-file summarization and automated knowledge base organization, and also supports users to integrate mainstream large models via Skillhub or API to build customized skill pipelines.

Currently, copilot is fully available on Mac, Windows, iOS, Android, and HarmonyOS platforms, and is being rolled out in batches through an application process. Amid intensifying competition in the AI assistant market, ima is trying to define the form of "next-generation knowledge management" by strengthening underlying memory assets and skill scalability. This iteration from simple information indexing to deep context awareness will significantly improve the delivery efficiency and digital asset accumulation capabilities of knowledge workers in complex tasks.