Enhancing interactivity between humans and large language models (LLMs), making them more accessible and aiming to bring the interface closer to the natural dynamics of human-to-human interaction.
Life OS enhances the UX of LLM interfaces by humanizing the chatbot experience. Users create personalized avatars, assign names, and engage in unique conversations with each "persona." These chatbots are individual, retaining memory only of the interactions they have with the specific user who engages with them.
The first step in improving products like ChatGPT is to move away from the traditional conversation-based organization model. Conversations should instead be treated as files within a humanized, grouped entity. This allows us to determine which "persona" remembers what, and we can create distinct environments for work, research, or personal discussions—just as we do in typical human relationships.
Additionally, users should have the option to create their own avatar for each "persona." This adds a layer of personalization and humanization, allowing users to further define the identity and role of each chatbot. Users will also be able to choose the associated language model, appearance, voice, and even provide a description of the personality that the chatbot should adopt.
Naturally, we need to visually engage the user and create subtle microinteractions that enhance the overall experience. It’s crucial that a project like this evokes a sense of security, as communicating the humanization of machines is no easy task. Ensuring users feel comfortable and confident while interacting with these avatars is essential for the success of such an initiative.
We’ve already seen initiatives that explore video calls using multimodal models, but now, with real-time 3D avatars, the opportunities are greater than ever. This opens the door to mixed-reality applications, similar to the "persona" concept in Vision Pro, as well as group calls featuring bot "actors" that can perform various tasks. One thing is clear: there’s no interface more familiar to us than a human face.