No. 01 · A companion for your collection

Know your shelf.
Find your next bottle.

Sillage reads the fragrances you own, shows what they say about your taste and where your shelf has gaps, then suggests what to try next, and explains every pick.

Try,,,,

or start your own, free

No sign-up for the sample: six bottles, ready in seconds, gone after 24 hours.

Light BlueAlienLa Vie Est BelleSauvageClub de Nuit Intense Man
AventusAcqua di GioCK OneJazz ClubAngel

Shelf note

Nothing here for a summer evening.

Fill gaps would start there.

The ten most-voted fragrances in the catalog, one per family.

The problem

Twelve bottles in, people still buy the wrong next one.

  1. 1

    A fifth fresh blue scent

    Without seeing the shelf as a whole, it's easy to keep buying the same profile under different names, and to miss that nothing covers a summer evening.

  2. 2

    Advice that doesn’t know you

    Databases know every perfume but not your collection. Chatbots recommend confidently, don't know what you own, and will invent a price if you ask.

  3. 3

    The right scent, the wrong bottle

    The cheapest-looking listing is often the EDT when you wanted the EDP, a 50 ml, a tester or a flanker with a similar name.

Why not just use…

Sillage connects the pieces the usual tools keep separate.

CapabilityFragrance databaseA chatbotA spreadsheetSillage
Knows what you own and how you rate it
Spots the gaps in your collection
Explains why a pick fits, with caveats
Can't invent a fragrance or a price
Checks prices for the exact bottle

How it works

I

Your shelf, mapped

Add what you own (typos are fine). Rate what you love; ownership alone counts for little. Sillage shows your scent mix, your signature accords and which seasons and times of day you can actually cover.

When you can wear them

Sample shelf
SpringSummerFallWinter
Day4422
Evening2244
Bottles suited to each season and time of day, from community votes.

II

Three ways to look for the next one

Similar, for more of what you love. Different, for a real change that still connects to your taste. Fill gaps, for what your shelf is missing. A deterministic engine ranks 3,000 well-reviewed candidates and every pick says why, and what's uncertain.

1

Clinique · Chypre

Aromatics Elixir

  • — Adds chypre (you own 0)
  • — Community votes favour it for winter
  • — Reviewers call it polarizing

78%

est. fit

An example pick for the sample shelf, with the engine's own reasons.

III

A price check for the exact bottle

Choose the concentration and size. Testers, flankers, wrong sizes and gift sets are set aside with the reason shown, and totals are only compared when shipping is known.

Looking for: EDP, 100 ml

Sample stores
  • Eau de Parfum, 100 mlExact match
  • Eau de Parfum, 100 mlExact match
  • Eau de Toilette, 100 mlDifferent concentration
  • Eau de Parfum, 100 ml, testerTester
  • Intense Eau de Parfum, 100 mlFlanker

The advisor

An agent with a job description, not a chatbot with opinions.

Some questions need several steps: read the shelf, decide which kind of search fits, run it, look closer at the best pick, maybe price it, then explain. That's what the advisor does. It's an AI agent (Claude or Gemini) that plans and calls Sillage's own tools. It never ranks fragrances itself and never touches your data without asking.

Reads
  • Your profile and ratings
  • The catalog
  • Engine results
  • Prices for one bottle
Proposes
  • Saving to your wishlist
  • Avoiding a note
  • A new budget

You confirm with a click

Never
  • Picks or ranks on its own
  • Cites what no tool returned
  • Quotes an unverified price
  • Buys anything

One agent, ten typed tools, at most six steps per answer. No model configured? It falls back to a rule-based mode that uses the same engine and says so. The full design

  1. You

    “A woody winter scent under $150 that isn't like anything I own?”

1 / 7
A worked example in the advisor's real format. In the app, every answer shows its own trace with latency and cost.

Measured, not claimed

Tested against cases written before the first run.

Every case and result
38,365
fragrances searchable
100%
precision matching the exact bottle (61 tricky listings)
29/32
shopper personas fully passing
0
invented products or prices reaching users

Two minutes with a sample shelf.