Product · E-commerce · Research

Zium

An energy and electrolyte gummies brand, built with AI, right up to the point of going into production.

In developmentJuly 2026 — ongoingFRUZEE

In short

Problem: Launching a supplement brand means going through market research, positioning, recipes, packaging, the store and the offer. A lot of very different jobs for a tiny team.
Solution: I relied on AI at every step to move faster, while keeping the decisions, and improved the product and the store version after version.
What it brings: A brand ready to be manufactured, with an online store built and reworked, in a short time.
Where it stands: Ready for manufacturing, but not on sale yet. The store and visuals exist and have gone through four versions.
My role: I led the project and made the positioning, range and design decisions. AI helped me with research, visuals and building the store.

The project

Zium is an energy and electrolyte gummies brand that we are developing within FRUZEE.

Launching it means going through steps that have nothing in common: understanding the market, choosing a positioning, deciding on recipes, designing packaging, preparing the discussion with a manufacturer, building an online store. In a tiny team, nobody is an expert in all of that.

The difficulty

Each step depends on the one before. If the positioning is vague, you cannot design the packaging. Without defined packaging, the conversation with the manufacturer goes nowhere. And if the range is unclear, so is the store.

How I went about it

I used AI at every step, while keeping the decisions.

For research, I mapped the market, compared existing brands and spotted the positioning still available. That is where I saved the most time.
For the product, I wrote up the concept, built the range and prepared the discussions with the manufacturer.
For visuals, I generated and retouched packaging ideas. That is where AI helped me least: a lot had to be redone.
For the store, I built it and improved it over four versions, up to adding a sales module.

Where it stands

Zium is ready for manufacturing, but nothing is sold yet. The packaging is still at mock-up stage, costs have not been checked against real quotes, and no customer has validated the positioning yet.

So this project does not show a commercial success. It shows that a brand can be taken from idea to manufacturing by a tiny team, doing very different jobs yourself.

Numbers

versions of the store

four versions of the theme, the last one with a sales module

Source · saved versions of the theme

What I did

Research
I studied the market, compared competing brands and spotted the positioning still available.
Product design
I defined the concept, the energy and electrolyte range, the formats, the recipes and the packaging direction.
Needs definition
I prepared the discussions with the manufacturer and listed the production constraints.
Assisted development
I built the online store theme and improved it over four versions.
Steering
I made the trade-offs between product, brand and sales channel.

Evidence

Four versions of the store
verified
I reworked the store four times, keeping each version. You can see the progress from one to the next, up to the sales module added in the last one.
Source
saved versions of the theme
Verify with
Compared the four versions of the theme and their dates.
A dated export of the store
verified
A full, dated export of the theme shows the real state of the store at that point.
Source
theme export dated 24 July 2026
Verify with
Read the export and its date.

Limitations

Nothing is on sale yet: no sales figures, no customer feedback, and a positioning the market has not validated yet.
The packaging is still at mock-up stage. The manufacturer spec is ready, but nothing has been produced.
Costs and margins have not yet been checked against real quotes.
At first, I kept the store versions as copies of folders. The theme is now in a versioned repository, but the early steps are harder to trace.
It is an ongoing project: the commercial choices are not set in stone.

What I take away

AI saved me a lot of time on research, comparing brands and positioning. Much less on visuals, which often had to be redone.
Keeping my versions as folder copies was a mistake. A versioned repository from day one would have cost me ten minutes and made every choice traceable.
Even with AI, the hardest part is still deciding, not producing.

What it could bring to a company

A product team can reuse the same approach (market study, positioning, concept, mock-up, sales page) to test a launch before committing resources.
A marketing team can compare several creative directions on a real page rather than on mock-ups.

Technologies

Shopify
Liquid
HTML
CSS
Image generation and editing tools