# Technology ### Visual Each A.I. model undergoes an intensive training process, whereby individual body parts, including the face, are repetitively fine-tuned throughout the course of multiple epochs. This ensures that the model will produce consistent results across a wide array of different poses, lighting, background environment, etc.

Demo training process using Margot Robbie as an example

After this process, we can generate an essentially infinite catalog of hyperrealistic image and video content that can be distributed across social media platforms such as Instagram, TikTok, OnlyFans, and more. For example, you can prompt the AI to show "\[Model name] standing in hotel room wearing lingerie"

[Model name] standing in hotel room wearing lingerie

Our AI models will never age, never fall out of shape, and can produce content 24/7/365 from anywhere in the world. Compare this to the tedious process that a real human model needs to go through in order to set up the photoshoot, travel to the location, assemble the wardrobe, hire a photographer, and then take hundreds if not thousands of photos in order to get the best looking one, not to mention simply maintaining enough motivation to continue to be in the business. You can begin to see why Lush's AI influencers are a 10x improvement over your typical human OnlyFans/Instagram/TikTok influencer.

Video generation from a still image using AI

{% file src="/files/Zu2NghbR0KAVvlJ4Iimg" %} Hyperrealistic video example {% endfile %} ### Speech While the visual component is the focal point of Lush's AI models, there are other important elements that go into creating the ultimate AI influencer, namely AI-powered engagement. As discussed in the [Rise of Influencers and OF](/lushai/lush-thesis/rise-of-influencers-and-of.md) section, it is already a common practice for OnlyFans models to outsource conversations with their fans to low cost virtual assistants, often located in lower income countries such as Vietnam or the Philippines. These VAs then manually run pre-planned chat scripts and send the fans pre-uploaded exclusive content. The inevitable next evolution is for specialized AI that is trained on intimate chat scripts to replace these VAs and fully automate the conversation process.

Who you are really talking to when you DM that OnlyFans model

Speech in the digital realm can take on multiple forms, including written text, audio, as well as video. Whatever the form, the underlying need is the same - an uncensorable and permissionless LLM that can be trained for the specialized purpose of engaging in the art of intimate conversation. Due to the sexualized nature of this use case, traditional LLMs such as OpenAI's ChatGPT are certainly not fit for the task, as it would go against their Terms of Use. This is where [our partner enqAI](https://enqAI.com) comes in. They are working on producing a uncensorable, permissionless, and blockchain native LLM that we can integrate into our AI companions to fully automate intimate conversations. Moreover, they have a working text-to-audio tool that is able to produce audio in any voice if provided a text input. This will give another life-like dimension to our AI companions. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://lushai.gitbook.io/lushai/lush-thesis/technology.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.