I use AI to help me blind buy perfumes, and it's actually perfect

I have a confession. Most of you know me for my obsession with phones, but I also have a secret obsession with perfumes. Over the past few years, I’ve built a collection, and most of them have been blind buys — perfumes I’ve bought without smelling them, based merely on their note pyramids and user reviews. It’s a fun gamble that more often than not ends up with me smelling nicer than ever, but I’ve also ended up with expensive bottles that I’ve barely touched again.

With AI infiltrating every facet of our digital lives, I wondered: Could Gemini help me find the perfect perfume for my next blind purchase? As it turns out, Gemini knows my nose better than I do.

Teaching Gemini my “nose”

My perfume collection is bordering on an obsession, fueled by a mountain of designer bottles and well-known clones that could easily last me through the decade.

As if that weren’t enough, I also subscribe to a monthly sampler from my local clone house. The data nerd in me has meticulously rated close to a hundred samplers in an Excel spreadsheet, along with my main collection in a separate sheet.

With this large dataset, Gemini had an excellent starting point for analyzing my scent profile.

When I fed this data into Gemini, I gave it a specific constraint: Weigh the main collection higher. I noted that the sampler data should be secondary because clones aren’t always 1:1 reflections of the original. This provided the AI with a massive yet nuanced starting point to understand my specific taste.

Letting AI decode my senses

The brilliance of using an LLM like Gemini is that it can cross-reference chemical compounds and user sentiment at a scale I simply cannot for such an extensive collection. While I see a “Fresh Summer Scent,” Gemini sees Calone, Ambroxan, and Citral — and it knows how those molecules correlate with my past ratings. What would have otherwise been hours of manual research and meticulous data organization is now condensed into a concise query that returns results in a few seconds.

By analyzing the commonalities across my ratings, Gemini was able to identify patterns I had missed. For instance, I know that I prefer “blue” perfumes — perfumes with an “aquatic” profile — but Gemini noted that I also have a clear affinity for “sweet” profiles and loud projections.

AI’s understanding of my nose was nuanced as well, as it could figure out that, despite my preference for “blue” perfumes, I consistently dislike “synthetic blues” (those sharp, metallic notes common in mass-market fragrances). I also dislike “earthy” notes, which I had previously mistaken for a dislike of “leathery” notes.

Not only did Gemini figure out what I like and don’t like, but it also figured out a major gap in my collection. AI noted that I have an excellent rotation of perfumes for daily wear in the summer heat, but I was missing spicy, dark fragrances that work the best for cold winter nights.

Thinking about it, it made perfect sense! I live in Mumbai, a tropical city that barely gets cold, so I had subconsciously avoided these heavier scents. However, every collection needs a “statement” fragrance for the occasional cool evening, and I had a blind spot the size of a spice rack.

AI vs hype: The ultimate “Buy or Pass” assistant

Armed with all of this data, Gemini now acts as my primary filter. Whenever my “perfume guy” posts a hot deal, I quickly ask the AI assistant for a verdict.

Recently, it told me to skip Hawas Rasasi Ice, despite the social media hype. Ordinarily, I would have jumped on the hype train, but Gemini told me I already owned a close competitor (Hawas Rasasi), and my historical data suggested I would find the specific synthetic notes within Rasasi Ice off-putting.

On the flip side, Gemini gave me the green light to buy two specific clones of Parfums de Marly scents: Lattafa Maahir Legacy (a clone of Sedley) and French Avenue Liquid Brun (a clone of Althair).

The AI pointed out that these two would fill gaps in my collection, with Maahir Legacy serving as a refined aquatic scent and Liquid Brun as a sweet-spicy winter scent. Both also align with my preference for high-performing, “loud” profiles, making them worthy of a place on my shelf.

I took the plunge. They are the best blind buys I’ve ever made, hitting the right spots without feeling redundant in my collection, or burning a hole in my pocket for that matter.

Gemini is a great shopping companion when you don’t know exactly what you want

It’s easy to know that you like something and you want more of it. However, it’s much more difficult to figure out why exactly you like something and why you dislike other things. Within my perfume collection, Gemini could find the deeper common notes that had escaped my eye in this mountain of surface-level data.

You don’t need a spreadsheet of 100 perfumes to make this work. If you’re starting out in building a perfume collection, you can simply tell Gemini or any other AI assistant: “I like these three perfumes, but I hate the smell of lavender and wet dirt. What should I buy next?” As you rate more scents and build and maintain more reference data, the recommendations only get sharper.

I wish AI had existed when I started off on this hobby. I could have saved a lot of money and shelf space by avoiding bottles that don’t suit my nose, and gotten a reality check on hype-driven FOMO.

It also vastly simplified research and comparison — I don’t need to spend hours dissecting a perfume if I can get a data-backed opinion in seconds. I can still choose to dive down the rabbit hole (it’s part of the fun, after all!), but now armed with better knowledge on how a scent would likely work out for me.

While I’ve used Gemini for my perfume collection, it can be used across hobbies. I’ve used Gemini to figure out an optimized “spending strategy” across my 10+ credit cards. I went to it for advice on how to tune my bike after upgrading to racing suspension, consulted it after making modifications, and got guidance on what to tweak next. I even used it to figure out my first IEMs to buy, given a bunch of my constraints (Gemini told me to go for the Meze Audio Alba ($19.99 at Amazon), and this one is even SoundGuys approved). The possibilities are endless for what you can do with AI when you feed the right data and ask the right questions.

The “Human-in-the-Loop” caveat: AI cannot replace you

Fragrances are inherently subjective — what smells good to you may not smell good to others around you, as everyone notices different notes differently. AI also cannot account for your specific skin chemistry or how a scent might evolve in your local humidity. So there’s still plenty of human subjectivity and other variables involved in the whole process of finding the right perfume.

I am not saying you should give AI the complete reins to your life, or even your hobbies. I still enjoy researching and reading about new perfumes, and I only go to Gemini when I need purchase advice. Even with that, I don’t blindly trust the recommendation — AI can and has been wrong, so I still follow my gut when taking the final decision.

At the end of the day, AI cannot replace you or your nose. But in a world of infinite choices and aggressive marketing, Gemini is the filter that helps me find the scents I actually want to wear.