How Skin AI Works

OVESSI Skin AI is a vision model trained to analyse skin appearance and recommend one of five layered rituals. You can upload a photograph or answer four simple questions. The system examines texture, tone, and visible patterns, then matches you to a ritual that begins with the thinnest layers first. This page explains what the model looks at, how it protects your privacy, and what it does not claim to do.

What the model analyses

The Skin AI vision model examines photographs for visible markers of skin behaviour. It looks at surface texture, evenness of tone, the presence of small variations in pigmentation, and the way light reflects from the skin. These observations help the system recognise patterns that correspond to different skin states. The model does not diagnose conditions. It does not identify medical concerns. It simply matches what it sees to one of five rituals, each built around a different layering sequence. If you choose the question path instead of uploading a photograph, the system uses your answers about texture, sensitivity, and current habits to arrive at the same type of match. Both routes lead to a single ritual recommendation. The model was trained on a wide set of images and text responses, but it remains a tool for guidance, not a clinical assessment. It works best when you treat the output as a starting suggestion, not a final instruction.

Input methods and outputs

You have two ways to interact with Skin AI. The first is to upload a photograph taken in natural light, with your face clean and unobstructed. The model processes the image on our servers, analyses the visible features, and returns a ritual match within seconds. The second method is to answer four questions about how your skin feels, how it responds to products, and what concerns you notice most. Both paths produce the same outcome: a matched ritual from our set of five. Each ritual is a four-step sequence that begins with the thinnest formulation and builds to the richest. The output includes the name of your ritual, a short description of why it suits you, and direct links to the products in that sequence. The system does not store your photograph after processing. It does not build a profile over time. It delivers one recommendation based on one input, then discards the image data. You can return and run the analysis again whenever your skin changes or you want to reconsider your approach.

Privacy and data handling

When you upload a photograph, the image is sent to our server for analysis and deleted immediately after the model returns a result. We do not save your photo. We do not use it to train future versions of the model. We do not share it with third parties. The same applies to your answers if you choose the question path: your responses are processed, matched to a ritual, and then discarded. No personal identifiers are attached to the input. The system does not know your name, email, or purchase history unless you choose to create an account separately. If you want to understand how we handle all data across the OVESSI site, you can read our full Privacy Policy. The model itself runs in a secure environment. The analysis happens quickly, quietly, and without leaving a trace. You remain in control of what you share and when.

What it is not

Skin AI is not a medical tool. It does not diagnose skin conditions, detect disease, or replace advice from a dermatologist. It does not claim to be perfectly accurate for every face or every lighting condition. The model makes educated guesses based on visible patterns, but it cannot account for factors it cannot see, such as allergies, hormonal shifts, or underlying health issues. It is a starting point, not a prescription. If you have persistent concerns about your skin, you should speak to a qualified professional. The ritual recommendation is a suggestion. You are free to adjust it, skip steps, or ignore it entirely. We built Skin AI to make the choice of ritual easier, not to impose a single correct answer.

Where to go next

If you want to try Skin AI yourself, visit the Skin Reading page and choose either the camera or question path. Once you receive your matched ritual, you can explore all five options on the All Rituals page to compare layering sequences and see how they differ. You can also read more about our approach to privacy in the Privacy Policy. Skin AI is one tool among many. Use it when it helps, set it aside when it does not.

Common questions

Does the model improve over time? The current version is fixed, but we may refine it in future releases based on aggregated, anonymous feedback.

Can I use the same photo more than once? Yes, but the model does not remember previous uploads, so each analysis is independent.

What if I disagree with the recommendation? You can choose a different ritual manually from the All Rituals page or adjust the suggested sequence to suit your preferences.