The ethical guidepost for the beauty industry.
An AI and 3D visualisation platform that lets people see realistic outcomes of non-invasive beauty treatments before they commit, and gives clinics an ethical, compliant way to run their operation.
Most people book non-invasive treatments without ever seeing a realistic outcome, and the few visualisation tools that exist are expensive, inaccurate and ethically careless. Clinics, meanwhile, juggle bookings, consent and compliance on tools never built for them.
I researched, designed and validated a dual-sided iOS product end to end: an AI and 3D pre-visualisation tool grounded in honesty for users, and a calm management system for clinics. Every decision was traced back to research and an explicit ethics framework.
Non-invasive beauty is one of the fastest growing markets in the world, yet the moment that matters most, deciding whether to go ahead, is left to guesswork, idealised social media and sales pressure.
I kept returning to one uncomfortable truth from the research: people are making irreversible, expensive decisions about their own faces with almost no reliable way to picture the result. When it goes wrong, the cost is not only financial. It is anxiety, time off work and lasting damage to self-image.
Source: SaveFace, the UK's government-approved register for medical aesthetic treatments.
Individuals need a way to visualise post-treatment outcomes, but this raises real concerns around realistic expectations, consent and emotional impact. The answer has to balance visualisation with ethics, for users and clinics alike.
Plenty of apps predict surgical outcomes. For non-surgical treatments there were only three serious players, BeautyFix, EntityMed and AEDIT. I reviewed each myself, shared them with friends and put them in front of two clinics, then scored them with a normative analysis.
The verdict was consistent: clunky interfaces, five to seven treatments at most, expensive, inconsistent accuracy, no real information and no ethical guardrails. A clear, untapped gap for something more accurate, more customisable and genuinely trustworthy.
Meet Lauren, 27, a professional dancer weighing up a treatment to lift her confidence. She is stuck: she will not commit without seeing how she would actually look, and that uncertainty drags her decision out and feeds her anxiety. She needs honesty, a realistic preview and information she can trust, not a filter that flatters.
On the other side sits the clinic owner, running a busy practice on a patchwork of tools. Bookings, client records, consent forms and finances all compete for attention, and a single dropped detail can cost trust or compliance. They need one calm, credible system that keeps the operation tidy and the data safe.
I ran the project end to end through a user-centred double diamond: diverge to understand, converge on the real problem, diverge again to explore, then converge on a validated solution. Every phase fed the next, and I grounded the design in established theory so the decisions could stand up to scrutiny.
Before drawing a single screen I went deep on the literature: how marketing and social media shape beauty decisions, where cosmetic procedures cross ethical lines, what the law demands of biometric and facial data, and how trust is actually built in a product.
That grounding changed the brief. This could not just be a clever visualiser. It had to actively protect people from the exact failures the data exposed, unrealistic expectations, poor consent and careless data handling, while still feeling premium and desirable.
Five user interviews, three expert interviews with doctors and cosmetic surgeons, and an anonymous survey. I ran a thematic analysis on everything and four clear themes surfaced. Each theme became a concrete design decision.
It would have been easy to chase only the upside. Instead I mapped the outcomes I did not want, then designed against them. A visualiser like this could deepen insecurity or be read as a promise rather than a simulation, so the guardrails became features, not afterthoughts.
I explored widely, from live AR scanning and smart mirrors to at-home devices, then scored every concept against a product design specification using a Pugh matrix. The winning direction combined Concept 1's freedom to explore the app with Concept 2's AI-driven accuracy. It scored a clear plus ten against the datum.
I had placed the calendar inside "My Clinic". Closed card sorting with five participants showed people instinctively expected it on the home screen. Rather than defend my structure, I rebuilt the information architecture, for both the user and clinic sides, around where people actually looked.
It is a small change on paper. It is the difference between an app that feels intuitive and one that quietly fights its user every day.
Research links high-end branding with perceived reliability, so the identity leans into quiet luxury. Couture sets a confident, high-fashion tone for headlines, SF Pro keeps the interface clean and highly legible. The palette is deliberately restrained.
I tested with seven users and four industry experts across two rounds each, using heuristic tasks, Likert scales and think-aloud. Here is the honest part: my first wireframes failed a usability test before they even tested the interface. People could not read my handwriting, so I lost the insight. I switched to mid-fidelity and the feedback got sharp again. Fidelity has to match the question you are asking.
I applied Mortier et al.'s Human Data Interaction framework across both sides of the product. For something handling faces and health data, this is what separates a trustworthy product from a liability.
People see what data is collected, how it is used and who can see it, before they sign up. The home dashboard simplifies complexity to avoid overload.
Users can access, edit, retake, share or delete their results and data at any time, with straightforward, GDPR-aligned controls.
Choices are revisitable as needs change. If someone is not comfortable, they can simply decline, no dark patterns, no traps.
Users pick a treatment, scan their face with clear guidance, and watch a realistic outcome unfold over time, from swelling to healed. An intensity slider keeps it grounded, and disclaimers keep it honest. This was the single most anticipated feature in every interview.
Source-cited information on every treatment, with wellness-led options highlighted and the reason for it made explicit. It answers the questions people were otherwise taking to Instagram, where 92% had never had a proper consultation.
An expandable calendar that opens from month to day, client directories, finance with clear month-on-month comparison, and a forms area where clinics can import and manage their own consent documents, a direct response to expert feedback.
A selection of final screens across onboarding, home, results and the clinic dashboard.
I costed and modelled the product as a real venture. A dual revenue stream keeps it sustainable: tiered clinic subscriptions and flexible scan packages for individuals, on top of a roughly £60,000 build, validated against development quotes.
"Looks high end and trustworthy. I was bought during my first impressions."
Industry expert, final clinic-side evaluation
Beyond the testing, the concept and business model were presented at an Industry Review Evening, where the direction was validated by industry voices including inventor and entrepreneur Tom Pellereau, whose feedback aligned closely with the themes in my own research.
The hardest tension in the whole project was this: people wanted deep customisation, but the more control you give, the easier it is to create an unrealistic expectation, the exact harm I set out to prevent.
Solving it meant restructuring the architecture around a transparent simulation tool, pairing every customisation with honest constraints, disclaimers and ethical nudges. Customisation and honesty were not allowed to trade off against each other. They had to hold hands.
Formosa · Mo Rezaei · Product & UX Design