Blue
Your face, understood.
AI skincare diagnosis — from a single selfie to a regimen built for your skin.
Product & Pitch Overview · 2026
THE PROBLEM
Skincare is guesswork — and it's expensive.
$300+
wasted per year
Shoppers buy by trend and packaging, not by what their skin actually needs.
70%
use the wrong actives
Mismatched ingredients trigger irritation, breakouts, and wasted spend.
1 in 3
have a sensitivity
Allergens and reactive ingredients hide in plain sight on labels.
Zero
real diagnosis in-aisle
A dermatologist visit is costly and slow; shelves offer no personalization.
2
THE SOLUTION
A dermatologist-grade skin read
in the time it takes to take a selfie.
LUMÉ turns one front-camera photo into a complete, personalized skincare regimen — analyzing your skin, accounting for your sensitivities, and recommending the exact products and the order to use them.
Instant
Personalized
Sensitivity-aware
No appointment
3
HOW IT WORKS
Five steps. One selfie. Zero guesswork.
1
Take a selfie
Guided front-camera capture with lighting checks.
›
2
Facial analysis
Computer vision maps 200+ facial zones.
›
3
Detect concerns
Identify acne, redness, pores, lines, texture, tone.
›
4
Add sensitivities
Flag allergens & ingredients to avoid.
›
5
Get your regimen
A ranked AM/PM routine matched to your skin.
4
STEP 1 — CAPTURE
1
No forms. No quizzes. Just look at your phone.
A selfie is the only input
Guided capture
On-screen framing, lighting, and distance checks ensure a clean, analyzable image every time.
Works on any phone
Standard front camera — no special hardware, attachments, or lab kit required.
Private by design
Images are processed securely and never sold; users control retention and deletion.
3 seconds to result
From shutter to analysis, the experience feels instant and effortless.
5
STEP 2 — ANALYSIS
2
Computer vision reads the skin the way a clinician's eye would.
Automatic facial analysis
200+ facial landmarks
We segment the face into zones — forehead, cheeks, T-zone, under-eye, jawline.
Multi-signal models
Texture, tone, oiliness, hydration, and pigmentation are scored per zone.
Calibrated for lighting
Color-correction normalizes for ambient light and camera differences.
Trained on diverse skin
Models validated across skin tones, ages, and conditions to reduce bias.
6
STEP 3 — CONCERNS
3
We translate pixels into the concerns people actually care about.
Recognition of skin concerns
Detects what matters
Acne, redness, enlarged pores, fine lines, dark spots, uneven texture & tone.
Severity scoring
Each concern is rated and ranked so the routine targets the biggest wins first.
Visual skin map
Users see exactly where each concern appears — on their own face.
Progress tracking
Re-scan over time to watch concerns improve and measure what's working.
7
STEP 4 — SENSITIVITIES
4
Personalization that protects, not just prescribes.
Input of product sensitivities
Flag known reactions
Fragrance, alcohol, retinoids, acids, essential oils, and specific allergens.
Smart ingredient filter
Every recommendation is screened against the user's avoid-list.
Pregnancy & condition safe
Surfaces flags for rosacea, eczema, and other context-specific needs.
Conflict detection
Warns when two products shouldn't be layered together.
8
STEP 5 — REGIMEN
5
A complete routine — products, order, and timing.
Personalized regimen
Ranked product matches
Specific products matched to concerns, budget, and sensitivities.
AM & PM routines
Step-by-step order of application, with frequency guidance.
Why each step
Plain-language reasoning so users trust and stick with the routine.
Reorder & restock
One tap to buy, with reminders timed to when product runs out.
9
THE DATA MOAT
Recommendations grounded in science
Every ingredient we recommend is backed by an automatically-built knowledge base. Our literature-ingestion pipeline reads a product's ingredient list and enriches each ingredient with chemical data and peer-reviewed evidence — no manual data entry.
Self-building database
Point it at any product label and the knowledge base grows itself.
Evidence-backed
Each ingredient links to PubMed studies and PubChem chemical descriptors.
Deduplicated & canonical
INCI names are normalized and matched so every compound is stored once.
Compounding moat
The more products ingested, the smarter every recommendation becomes.
10
AUTOMATIC LITERATURE INGESTION
From an ingredient list to a living database
1
Parse INCI list
Split & normalize the raw ingredient declaration into ordered, deduped ingredients.
›
2
Resolve compound
Match each ingredient to a canonical compound — or create it — via INCI + aliases.
›
3
Enrich
Pull INCI properties, PubChem chemical descriptors, and PubMed research articles.
›
4
Status & store
Track enrichment state (complete / partial / review) and persist the linked graph.
Built today — a Django pipeline (formulation_ingest) that turns any product label into structured, evidence-linked ingredient data, fully tested.
15
UNDER THE HOOD
The intelligence behind the glow
Vision
Facial segmentation & dermatological feature detection on-device + cloud.
Knowledge
Ingredient graph linking 10k+ products to actives, concerns & conflicts.
Matching
Ranking engine scores products against the user's skin profile & avoid-list.
Reasoning
LLM generates clear, personalized explanations and routine guidance.
16
MARKET OPPORTUNITY
A massive market hungry for personalization
$190B
Global skincare market
$13B+
Beauty-tech & diagnostics
28%
Annual growth in AI beauty
Consumers already shop for skincare online and crave guidance. LUMÉ captures intent at the exact moment of decision — turning diagnosis into purchase.
11
BUSINESS MODEL
Three ways to monetize one selfie
01
Affiliate & commerce
Revenue share on recommended products users buy in-app.
02
Premium subscription
Progress tracking, unlimited re-scans, and pro routines.
03
Brand partnerships
Sponsored placement and anonymized skin-trend insights for brands.
12
WHY WE WIN
Diagnosis + sensitivity + commerce — in one flow
Capability
LUMÉ
Typical apps
Selfie-based diagnosis
✓
Quizzes only
Sensitivity-aware filtering
✓
Generic suggestions
Per-zone severity mapping
✓
One-size scores
Routine order & reasoning
✓
Product lists
Progress re-scans over time
✓
No follow-up
13
ROADMAP
Where we're headed
NOW
Core flow live
Selfie → analysis → concerns → sensitivities → regimen.
NEXT
Tracking & commerce
Progress re-scans, in-app checkout, restock reminders.
SOON
Body & brands
Expand beyond face; brand partner portal & insights.
FUTURE
Clinical-grade
Derm partnerships, telehealth referrals, predictive skin health.
14
Skincare that finally
knows your skin.
LUMÉ — Your face, understood.
Let's build the most personal skincare experience in the world.
weilan@chariotmove.com