Everyone is showing off their Twitter earnings, and I am very anxious, very anxious, but I am helpless.

My Stripe account shows a problem, and I still don't know how to handle it. I actually received a payout from Twitter once, but afterwards it just showed Stripe issues. I’ve tried countless times to communicate with the premium official account, but every time I get cold, robotic responses from AI. It keeps repeating mechanical answers that have nothing to do with my questions.

And every time I ask, I have to restate my question to it. It really confuses me. I no longer want to contact the official team. I believe everyone has encountered this situation in daily life, such as calling the bank—AI always spouts a bunch of nonsense, and in the end, you still have to talk to a human customer service representative.

AI claims that its development is now very mature. Can't AI have real memory, automatically store and retrieve user information? Can't it achieve more human-like conversations? Until I saw EPHYRA @EPHYRA_AI.

EPHYRA aims to create an AI companion that supports emotional, memory, and motivation systems, allowing virtual characters to think, grow, and express personality like humans. Recently, the system has made key breakthroughs in four core areas: perception, emotion and cognition, behavior and expression, and self-model.

EPHYRA adds an "external memory bus" at the bottom layer, enabling the model to store important information you mention on an external hard drive. In the future, your chat records will be continuously stored in a database, and the model can recall them within milliseconds, truly achieving memory-enabled conversations.

Previously, AI was passive, mostly requiring humans to ask questions proactively. But when humans are in a bad mood and don't want to talk, EPHYRA’s AI model adopts a motivation thinking system, allowing it to perceive human emotions within certain thresholds and reach a level of proactive communication.

Even more interesting, the AI model created by EPHYRA will have genuine emotions. The AI model is equipped with an affinity system and emotional feedback mechanism. If you upset it, it might become moody or even need your comfort to get better. Isn’t that an excellent level of authenticity?

I think such an intelligent AI model must be very expensive, and the possibility of owning one in the future seems limited. But according to EPHYRA’s explanation, they have already built a matrix including large, medium, and small models, from simple to complex, achieving "automatic model download and management." In the future, every user might be able to "afford" their own AI companion. I’m still a bit hopeful.