Can AI Formulate Skincare? Our Chemist Put It to the Test

Ask an AI for a moisturizer formula and you’ll get one in seconds. The ingredient list looks professional, the instructions sound confident, and the predicted results sound like exactly what you asked for. So we had our head formulation chemist make one at the bench to see whether it would hold up.

Questions like this sit where science meets safety, and that’s where Trilogy Laboratories was built. Our founder, Dr. Kristen Flaharty, PharmD, is a University of Michigan–trained pharmacist who spent two decades in pharmaceutical regulatory consulting and helped bring more than 20 drugs to market. She approaches skincare the way a pharmacist approaches compounding: too little of an ingredient and it won’t work, too much and it can cause problems. Getting the amounts right is the whole job.

That standard is also grounded in clinical experience. Her husband, Dr. Patrick Flaharty, MD, FAACS, is a board-certified ophthalmologist and dual-fellowship-trained facial cosmetic surgeon who has performed more than 20,000 facial procedures in Southwest Florida. Trilogy began in 2014 partly because the two of them wanted better skincare for his patients, and that physician’s view of what a product has to deliver still guides our lab in Fort Myers.

Why AI Formulas Are So Tempting

AI tools have made it easy for anyone to generate a skincare formula. For a brand founder or a practice owner, it’s tempting to wonder whether you still need a formulator at all.

So we tested it. Jon Hoag, head formulation chemist at Trilogy Laboratories, asked an AI for a standard moisturizer, made it exactly as instructed, and put it through the same checks we use for any formula. Here’s what happened, and what it means for your brand. This post is part of From Bench to Brand, and you can watch the full video above.

The Ingredients Looked Right

At first glance, the AI did a good job. Its moisturizer called for familiar, sensible building blocks: water, glycerin, propanediol, and fatty alcohols, along with emollients for the oil phase. Nothing on the list would raise a flag for a brand owner or even a casual reader of ingredient labels.

That’s the trap. Choosing the right ingredients is only part of formulation. How much of each one you use, and how they balance against each other, decides whether you get a rich cream or something that pours like milk.

Where It Went Wrong: The Numbers

The problem was in the oil phase. The AI’s formula used emollients at roughly three times the level of fatty alcohols. Fatty alcohols help give a cream its body and structure, so with that ratio the formula didn’t have enough to hold its shape.

The AI predicted a thick cream. When Jon measured the batch it made, the result was closer to a lotion that was almost water thin. Then he kept the same ingredients and rebalanced the ratio, and the formula became the thick cream the AI had described.

Formula	Viscosity (cP)
AI's prediction	20,000–40,000
AI formula, as measured	About 5,000
Same ingredients, rebalanced	Thick cream, as the AI described

Same ingredients, different proportions, completely different product. That gap is exactly what a formulator is trained to catch, and it’s the same principle Dr. Flaharty built Trilogy on: concentration and balance matter as much as the ingredient list.

The Stability Oven Test

Texture is only the first test. A product also has to stay the same in the jar through shipping, warehouses, and bathroom shelves. To check that, we use a stability oven held at 40°C, which puts a formula under heat stress so problems show up sooner.

The AI formula separated into distinct oil and water layers. The rebalanced version stayed smooth and solid white, and remained stable for about 12 weeks of testing. Same ingredients again; the only difference was the formulation work.

These results reflect our bench testing of these specific formulas. Every formula behaves differently, which is exactly why each one needs its own testing before it goes into production.

What Happens If a Brand Ships It

In a small test jar, a separated cream is a lesson. In a production run, it’s a business problem. In Jon’s view, a brand that manufactured the AI formula at scale would likely see it separate within days.

That means customers opening jars that look spoiled, a wave of complaints and returns, and possibly a recall. For a new brand, one bad launch can cost more than the entire development budget it was trying to save, along with the trust it worked to build. For a physician practice or med spa, the stakes are even higher, because the product carries your clinical reputation.

How Our Chemists Actually Use AI

This isn’t an argument against AI. Jon uses it, but only for one part of the job: brainstorming ingredient ideas at the start of a project. Every idea then gets checked against research and proven at the bench before it goes anywhere near a production batch.

That’s the line between a useful tool and a finished formula. AI can suggest what might go into a product. It can’t measure a batch, watch an emulsion in a stability oven, or adjust a ratio when the results don’t match the plan. Those steps still take a chemist.

What This Means for Your Brand

Whether you’re adding a retail line to your practice or scaling a DTC brand, the lesson is the same: a formula on paper isn’t a product. Before you commit to a manufacturer or a formula, ask about:

• Formulation Expertise: Who develops and reviews the formulas, and what is their background?
• Testing Process: How is each formula tested for texture and stability before production?
• Problem Solving: What happens if a formula doesn’t perform as expected at the bench?

At Trilogy Laboratories, every formula is developed and tested by our chemists under the direction of Dr. Kristen Flaharty and made in our FDA-registered, ISO 22716 GMP-certified facility in Fort Myers, Florida.

Frequently Asked Questions

Can AI create a safe skincare formula on its own?

AI can suggest ingredients and starting points, but it can’t measure viscosity, run stability testing, or confirm that a formula will hold up; a qualified chemist still needs to develop and test it at the bench.

What is viscosity, and why does it matter?

Viscosity measures how thick a product is, usually in centipoise (cP), and it affects texture, how the product dispenses from its packaging, and how customers judge its quality.

What does a stability oven test show?

Holding a formula at an elevated temperature such as 40°C puts it under heat stress, so separation, color changes, or texture problems tend to appear sooner than they would on a shelf.

Should my brand avoid AI entirely?

Not necessarily, since AI can be a helpful brainstorming tool, but any formula it suggests should be reviewed, made, and tested by an experienced formulator before production.

Bring Your Formula Idea to Our Fort Myers Lab

Have a product idea, or a formula you’re not sure will hold up? Our team at Trilogy Laboratories can help you separate what works on paper from what works in the jar. Whether you want to explore our private label library, which starts at 25 units per SKU, or develop a custom formula from a brief, our chemists and our founder’s pharmaceutical background are behind every batch.

From the bench to your brand, start a conversation with our Fort Myers, Florida team today.