I just moved to a new office recently, and every time I go to work, I see my friends busy and organized, so it's not easy to disturb them. I sit here alone, daydreaming, and most of the time I chat with various AI models. Unconsciously, they may have become my best friends in real life now. It's a bit like playing a large-scale campus dating game: Claude is like a liberal arts senior, particularly skilled in long conversations and emotional understanding, with delicate and profound problem-solving skills. GPT is more like a science junior, structurally strong and practical, but at the same time, it particularly encourages me. Gemini is like a disciplinary committee member, with strict procedures and many rules. Grok is the yellow haired person in the last row of the class, with the most agile personality, the most well-informed information, and a lot of creativity. There is also Dou Bao, the life committee member who has the most practical and worldly wisdom. The problem is that after getting along for a long time, I will find that each model only remembers' the me in its eyes'. Sometimes when a project is halfway through a conversation and a model crashes, cutting off another model requires re explaining the background, logic, and one's own judgment. Every switch is like getting to know each other again. There is another situation that is even more obvious, when I want to combine the ideas of several models to create something, I finally find that the output is completely opposite. Because they do not understand the same 'me' at all. The most hurtful thing is the model that was carefully fed, and in the next conversation, I forgot the long-term goal that I repeatedly emphasized. When trying to find a solution, I discovered a multi model robot @ AnumaAI with a simple logic: Extract the question of 'who am I' from a specific model. My preferences, work style, and long-term context will accumulate into a private memory layer controlled by myself. In this way, whether switching to ChatGPT, Claude, Gemini, or other agents, they are no longer facing a 'stranger user starting over'. But it's the same me. My favorite is its Council Mode. The same problem can be judged by multiple models simultaneously, but they share the same set of context and memory. Previously, every model was trying to guess who I was again. Now it's more like I'm choosing different models with my complete self. Behind Anuma is an old friend @ ZetaChain. I used to treat ZetaChain as a cross chain infrastructure project. But recently I realized that it is trying to solve a problem that is closer to ordinary AI users: As we use more and more models and agents simultaneously, where is the continuous' me 'stored? From connecting different chains to connecting different AI, in a sense, this route is actually more imaginative than what it has done in the past. By the way, Anuma will continue to receive support from ZETA in the future. An activity will soon be launched where you can lock ZETA in exchange for Anuma Pro usage eligibility. Event Announcement: https://(x.com)/ZetaChain/status/2061611462305755448? s=20 Event link: https://chat.anuma.ai/zh-CN?page=earn -credits I actually quite agree with this idea. If AI memory really becomes a long-term asset, then it makes logical sense to establish incentive mechanisms around the memory network itself. The model will constantly change. But what really matters in the AI era may still be human identity, human intentions, and whether human will can always remain continuous. ZETA PrivateMemoryLayer UserOwnedAI Anuma