Most fragrance advice starts at the surface: notes copied from a box, star ratings averaged into a number. Scenday Intelligence starts beneath the notes, at the chemistry of scent itself, and climbs. Each layer of understanding unlocks the next. This page is the whole ascent.
Scenday Intelligence is an intelligent system, purpose-built for fragrance and developed in-house with modern AI, including large language models. You can simply talk to it in your own words, or just use it: it builds your scent profile from whatever you give it, your words, your choices, your wears. There are no forms to fill and no editors writing its verdicts. What follows is what it understands, and in what order.
Layer I · The Chemistry
The foundation. What a perfume actually is, beneath the words.
§ 1
Beneath the notes
A note is a word. Chemistry is the thing itself.
Open any fragrance listing and you will find a pyramid: top notes, heart notes, base notes. It reads like a menu. It is actually physics. A perfume is a solution of aromatic materials, from a few dozen to a few hundred in one bottle, and every material evaporates at its own speed. The light molecules leave first: citrus peels, herbs, the bright rush of the opening. The heavy ones hold: woods, resins, musks, the part still on your wrist at midnight.
The pyramid is a timetable. Perfumery has ranked materials by evaporation speed for a century, since W. A. Poucher first classified them in 1923, and that property, volatility, is still the spine of how a composition is read.
Watch it in any composition. A grapefruit note is loud for twenty minutes and gone within the hour. A vetiver is quiet at noon and still present at dinner. Citrus and herbs burn through the opening; florals and spices carry the middle hours; woods, resins and musks hold the floor into the night. None of this is printed on a box. All of it is chemistry.
A note list tells you what a perfume contains. Volatility tells you when. Scenday Intelligence reads every fragrance the second way: as an event in time, not a list of words.
§ 2
Ingredients and dosage
What you smell, what is in the bottle, and how much of it.
Notes and ingredients are not the same thing. There is no leather in a leather note. Leather is an accord: an effect built from other materials, the way a chord is built from separate strings. Reading a fragrance means keeping three layers apart: the ingredients in the bottle, the notes you actually smell, and the accords they conspire to form.
It also means dropping a prejudice. Naturals are not automatically noble, and synthetics are not shortcuts. Materials like Hedione and Iso E Super carry much of modern perfumery, while some naturals vary harvest to harvest the way wine does. The intelligence treats both as what they are: materials with behavior.
Concentration is dosage, not prestige. Eau de toilette, eau de parfum, extrait: the labels mark how much aromatic material is dissolved in the bottle, and sometimes the formula itself changes between strengths. The same name at two concentrations can be two different perfumes. Scenday Intelligence reads each strength as its own composition.
The materials themselves are governed. The International Fragrance Association sets usage standards for hundreds of materials, and a serious reading of a fragrance is made knowing them.
The hour hand follows your reading: as it crosses the ladder, the light families go quiet and the heavy ones hold.
How much aromatic material is dissolved in the bottle. The same name at two concentrations can be two different perfumes.
Unlock
Once a fragrance is understood as chemistry in time, it can be described precisely. Precisely enough to be given a DNA.
Layer II · The Scent DNA
The fragrance genome. Description made exact.
§ 3
A DNA for every fragrance
Six axes, drawn by contract.
From the chemistry, Scenday Intelligence describes every fragrance in its own vocabulary: more than 10,000 notes and accords, and growing. Not from advertising copy. Not from a reviewer's mood. From the composition and how it behaves.
That resolution is too fine for a page. So this site draws every profile the same simple way: six axes, readable in one glance. The drawing is the site's convention. The vocabulary underneath is the intelligence's.
One sentence, read. This is the part the large language models do: natural language in, structure out.
In words: sentences become structured entries, filed into the levels above. Nothing stays prose; everything becomes checkable structure.
A reading keeps its uncertainty. A note that registers faint stays on the reading as faint: prominence two of five, written down as exactly that. Faint is not absent, and rounding it away would be a small lie.
A DNA is a description, not a score. No axis is better than another. A quiet citrus and a loud amber are equals on this page; the judging comes later, and by a different contract.
§ 4
One map of all scent
Every description becomes a position.
Describe one fragrance this way and you have a fact. Describe them all the same way and you have a geometry: every fragrance a position, every similarity a distance. Every composition stands in its own quarter of the map, its nearest neighbours named and ranked.
More than 120,000 fragrances are on the map, and the map is never finished. New releases join it, and each arrival sharpens the neighbourhoods around it. A census of scent would go stale in a season. A map only gets better.
In words: nearness on the map is computed from what two compositions share and how alike they behave. Geometry, not opinion.
| Level | What it holds | For instance |
|---|---|---|
| Materials | What is physically in the bottle, and at what dose. | naturals, synthetics, concentration |
| Notes | What you smell, named from the vocabulary: 10,000+ and growing. | grapefruit, vetiver, benzoin, flint |
| Accords | How notes conspire into effects none of them makes alone. | chypre, leather, gourmand |
| Behavior | How the composition lives over hours. | longevity, projection, the arc |
| Profile | The drawn summary a person can read at a glance. | six axes, as this site draws it |
Every level, every fragrance.
| What is written | Where it files |
|---|---|
| "bright grapefruit opening" | Notes: grapefruit · Behavior: a loud open |
| "gone by lunch" | Behavior: longevity, short |
| "too loud for the office" | Behavior: projection, high · Context: occasion |
Unlock
A single fragrance, described precisely, is a fact. More than a hundred thousand, described the same way, are a geometry. Geometry is what the deeper modules are built on.
Layer III · The Modules
What deep description makes possible.
§ 5 · Module
Behavior
How a fragrance lives over hours.
The first module predicts the wear. From composition and concentration, the intelligence reads a fragrance's arc: how loudly it opens, when it settles, how long it holds, how far it reaches. Longevity and projection stop being guesses pooled from strangers and become properties read from the thing itself.
Arcs differ by build. A citrus-forward composition opens loud and burns out by lunch. An amber barely projects and never quite leaves. A woody citrus opens bright and bitter, turns peppered and mineral, and holds a floor that outlasts the day, on most skin. That phrase, on most skin, is doing honest work. Layer IV is about the exceptions.
In words: how strongly a composition is present at any hour is computed from what is in it and how fast each part leaves. Longevity is where the curve meets the floor; projection is how high it opens.
§ 6 · Module
Kinship
Similarity, comparison, and what layers with what.
The second module puts the map to work. Nearness in scent space is kinship: it is how a comparison can be exact instead of vibes, how close to this, drier, quieter becomes a measurable sentence rather than a shrug. It is also how layering stops being luck: two DNAs can be read together, their overlaps and collisions named, before anything touches skin.
This module is already public across the site. Every comparison page, every duet, every list of neighbours is kinship at work, and each one cites the DNA it was computed from.
§ 7 · Module
Judgment
Verdicts by contract, sealed against standards.
The third module judges, and it judges in the open. Every verdict is scored by a published contract: five criteria, weighted 40 / 20 / 15 / 15 / 10, itemized so the arithmetic can be checked. One rule cannot be outweighed: no fragrance scores above 79 without the longevity to show for it. Strength elsewhere cannot buy the difference. A formula that can be gamed is not a standard.
The registers on this site are this module's published output: Scenday Intelligence writes them. Every judgment is gazetted: numbered, dated, stamped, and kept. The record is gap-free by design: 37 notices at the last reading, 23 of them judgments, and growing, every serial accounted for, nothing quietly removed.
In words: five criteria, fixed weights, and one constraint no strength elsewhere can buy past.
Each one cites the DNA it was computed from.
Itemized so the arithmetic can be checked. A formula that can be gamed is not a standard.
Unlock
Behavior, kinship, judgment: all of it is about the bottle. The bottle is half the story. The other half is the person wearing it, and the day they wear it on.
Layer IV · The Human and the World
The variables the bottle cannot know.
§ 8
The world's variables
The same bottle, in July and January.
Scent is carried by air, and air has weather. Heat speeds evaporation: a fragrance projects louder and dies sooner on a hot day. Cold flattens it to a whisper close to the skin. Humidity carries scent farther; dry air lets it fall. Season, climate, even the difference between a heated office and a winter street change what a perfume does.
A bright woody citrus leans sharp, loud, and short in a heatwave. At four degrees it turns quiet and mineral, and lasts. Same bottle. Different perfume.
Scenday Intelligence treats the world as an input. Where you live and what season it is are part of the reading, not trivia beside it.
In words: weather is inside the calculation, not a caption under it.
§ 9
The human variables
The last ingredient is you.
Skin is chemistry too. Its warmth, its oils, its dryness bend a composition. This is why your friend's compliment magnet dies on you in an hour, and why a paper tester strip is a rumor, not a verdict. The strip has no chemistry. You do.
Preference enters in your own words, when you feel like giving them. Tell the intelligence what you want: something woody but quieter, for the office, in a cold city. It reads sentences the way a good fitter reads a fitting. But talking is optional: pick, wear, and report, and it learns you just the same.
And life enters last: what you wear it for. An office has rules a Saturday does not. The intelligence reads occasion the way it reads weather: as a condition of the wear, not a footnote.
§ 10
The match, alive
A prediction, not a popularity contest.
A match is not people who liked X also liked Y. It is a prediction of an experience: this fragrance, on your skin, in your weather, for your evening. When the fragrance and the wearer are described in the same language, the prediction can be scored in the open.
In words: a match score is a weighted reading of five fits between one person and one composition, under the day's conditions.
Every point is accounted for. If you disagree with a 92, you can find the exact line you disagree with. That is the difference between a score and an opinion.
A condition of the wear, not a footnote.
| Member | F1 | F2 | F3 | F4 |
|---|---|---|---|---|
| M1 | 61 | 92 | 48 | 75 |
| M2 | 83 | 39 | 71 | 55 |
| M3 | 44 | 68 | 80 | 62 |
Match scores, one row per member: an illustrative grid. One row is one member scored against the field. The bordered cell is the worked match below, itemized.
| Component | Score | Out of |
|---|---|---|
| Profile alignment | 46 | 50 |
| Longevity fit | 14 | 15 |
| Occasion fit | 13 | 15 |
| Season fit | 9 | 10 |
| Intensity fit | 10 | 10 |
| Match | 92 | 100 |
Every point is attributable: score as layered evidence.
Unlock
A match this specific is already rare. It is also still a snapshot: the best prediction available on the day it was made. The frontier is what happens after you wear it.
Layer V · The Frontier
A Scent DNA unique to you, and moving.
§ 11
Every wear teaches it
Data no questionnaire could ask for.
Scenday Intelligence is built to learn from wear reports. Loved it. Too loud for the office. Gone by noon. Perfect in the cold. Each report is worth more than any quiz answer, because it is not an opinion about taste in general. It is a recorded fact about one fragrance, on your skin, in your weather, on a day that actually happened.
§ 12
The DNA moves
Version one is a sketch. Version nine is a portrait.
Your first Scent DNA is version one: drawn from the little you have told it, honest about how little that is. It is built to move. A wear that surprises it moves an axis. A season of wears redraws the neighbourhood it looks in. The system is designed so that two members who start from the same place will hold different DNA within a season of wearing, because no two people wear the same year.
In words: each wear moves the profile by exactly how much it surprised the current version. No surprise, no movement. A season of small surprises is a portrait.
§ 13
Unique by construction
Grown on your wears, not prompted out of you.
This is the frontier the layers were climbing toward: a Scent DNA that describes you and no one else. Not because you filled in a longer form, but because it was grown on your actual wears, and no one else lived them. It cannot be copied, and it cannot be faked.
Scenday Intelligence was built in-house with modern AI, including large language models, and it is still climbing. Every layer on this page is versioned, and none of them is finished.
Six axes drifting across versions, v1 to v9. The drawing is the site's convention; the resolution underneath is the intelligence's.
No surprise, no movement. A season of small surprises is a portrait.
Layer VI · The Ledger
What binds the intelligence: the contrast, the refusals, the record.
§ 14
Surface and depth
What a chatbot does with the same question.
You can talk to Scenday Intelligence the way you would talk to a chatbot, and the resemblance ends there. A chatbot is its conversation: text about text, gone when the tab closes. Scenday Intelligence is a system that happens to speak. Beneath any conversation, and there when you skip one, are the layers this page just climbed.
| The question | A general chatbot | Scenday Intelligence |
|---|---|---|
| What is this perfume? | A list of notes echoed from marketing copy, different at each retelling. | Chemistry in time: materials, dosage, and an arc across hours, on the record. |
| What will it do on me? | A guess, and the same guess for everyone. | A prediction from your skin, your weather, your occasion, scored in the open. |
| Does it know me? | It starts from zero every conversation. | A Scent DNA built to move with every wear, version by version. |
| Who is accountable? | No record, no standards, no serial. | Itemized scores, versioned standards, a gap-free public record. |
None of this makes a chatbot useless. It makes it shallow here. Fragrance is chemistry on skin in weather over time, and text about text cannot reach it. The conversation is only the doorway. The difference is what is behind the door.
§ 15
What it refuses, and where it is weak
The refusals are part of the method.
- On the fragrance side No sponsored placements, no brand payments, no hype weighting. A fragrance cannot buy an axis of its DNA or a point of its verdict.
- On your side Your wear reports shape your Scent DNA and nothing else. They are not sold, and they are not shared.
- Currently marked weak: projection The evidence behind projection readings is thinner than we want, especially for niche releases. The criterion is under revision for standards v2.1. We would rather flag it than pretend.
- Standards, versioned v1.0, 21 July 2026. v2.0, 28 July 2026. v2.1, in revision. When a standard changes, the version printed on every seal says so.
§ 16
Questions, answered
Is Scenday Intelligence a chatbot?
No, though you can talk to it. Chat is an interface; the intelligence is the system behind it. A general chatbot has read text about perfume and is nothing beyond the conversation. Scenday Intelligence holds a vocabulary of more than 10,000 notes and accords, and growing, a map of fragrances, a model of behavior on skin and in weather, and your evolving profile, whether you chat with it or never say a word. It can talk like a chatbot. It knows like a laboratory.
Does a human review the verdicts?
No. We build the intelligence and write its standards; Scenday Intelligence writes the verdicts. Every verdict is sealed against a printed standards version, and the record is gap-free, so nothing can be quietly rewritten.
What is a Scent DNA?
Two things share the name, deliberately. Every fragrance has one: a description in the intelligence's vocabulary of notes and accords, drawn from its chemistry. And every member has one: a profile of what they reach for, begun from whatever they first give it and built to move with every wear. A match is the two read together, under your weather and on your skin. The six-axis drawings across this site are the simple way we draw both.
Why does the same perfume smell different on my friend?
Because skin is an ingredient. Warmth, oils, and dryness bend a composition, so one bottle is a slightly different perfume on every person who wears it. That is not a flaw in the perfume. It is the reason a recommendation must know something about you.
How is this different from asking ChatGPT?
Depth of ground. A general chatbot has read what people wrote about perfume. It holds no model of chemistry, no map of scent, no record of you, and its answer changes with each retelling. Scenday Intelligence is built from the chemistry up, scores by published contract, and is designed to remember what it learns from your wears.
How does my Scent DNA become unique?
Through wears, not forms. Two members can start from the same place and hold near-identical first profiles. From there the paths split: every wear report moves the profile it belongs to, and no two people wear the same year. Uniqueness is not promised. It is constructed.
Why is the quiz on this site so short?
Because it is a sketch, on purpose. Five questions in about sixty seconds let you feel the method before the app. Real profiling happens inside the app, where the intelligence works with its full vocabulary and learns from your words, your choices, and your wears. The quiz starts the portrait. It was never meant to finish it.