The Fraghab story

I’m Alex Martin, a full-stack engineer in Michigan and a father of three. For most of the last decade I built businesses first and wrote software second. Fraghab is what happened when those two lives finally merged, and this post is the whole story of it. Fair warning, it’s a long one.

In 2017 I was one of the original founders of Lemonati Family Farms, a licensed cannabis cultivation facility with 20,000 square feet of canopy. I ran operations end to end, everything from compliance and staffing to supply chain and sales, and helped grow it into an award-winning brand generating millions a year in revenue. Nothing teaches you systems thinking like regulated agriculture. Every gram gets tracked, every process gets audited, and mistakes are expensive.

I left Lemonati to take the Operations Director role at Viola, the multi-state cannabis company that NBA veteran Al Harrington founded and named after his grandmother, whose glaucoma first showed him what the plant could do. Allen Iverson famously partners with the brand. I spent a year there running operations at a scale the farm never approached, with more markets, more facilities, and a lot more moving pieces. It was a different education entirely.

After Viola I went independent and consulted for operators around the industry. Everywhere I went I noticed the same thing: the software these businesses ran on was generic, rented, and fit them badly. So I started building custom tools for my clients, and somewhere in those years I stopped being an operator who codes and became an engineer with an operator’s instincts.

Today I’m the software engineer and marketing director at Planted Provisioning, a three-store dispensary group best known for a massive delivery footprint that covers the entire east side of Michigan. For them I built a completely bespoke e-commerce web application that has processed over $15 million in sales, plus a dedicated company ERP suite. All of it runs on a network of physical servers I built myself, housed here in Michigan.

That software and infrastructure touches every aspect of Planted’s operations and customer experience, from the first tap on the storefront to the systems running behind the counter. It’s one of the only consumer-to-business operations solutions in the cannabis industry. The marketing that keeps those stores busy runs through me too.

My daily languages are Go, TypeScript, Python, and Rust. Around them sit Next.js and React on the front, Postgres, SQLite, and Turso underneath, Docker and Redis in the middle, and lately the Claude API and MCP in more of my work every month. I care about the whole line from UX to deployment, mostly because in every job I’ve had, I was the whole line.

Outside of work I collect fragrances. A few bottles became a shelf, the shelf became a cabinet, and at some point I realized I was managing the whole thing from memory. Which decants were running low. What I wore to the last wedding. Whether that winter oud ever got a fair shot in the rotation. My head held all of it, right up until it couldn’t.

My answer, like everyone’s, was a spreadsheet. Name, size, price. The problem is that nothing which actually matters about a fragrance fits in a cell. How it wears in the heat, how long it lasts on skin, where you spray it, the night it earned its place in the rotation. A spreadsheet stores the label and loses the bottle.

The alternatives had their own problems. Discovery apps treat your collection as a byproduct of browsing, since their whole job is keeping you focused on the next bottle instead of the ones you own. Generic inventory tools don’t speak the language at all. No notes, no accords, no concentrations.

In March 2026 I stopped waiting for someone else to fix it and started building, for an audience of exactly one. The first version modeled every bottle as a complete record, logged daily wearings with spray counts, and decremented the remaining milliliters so the collection on screen stayed true to the shelf. There was no pitch deck and no roadmap. It was just the tool I had wanted for years.

One conviction shaped everything after it: the software should feel worthy of the category. Fragrance is precise and deeply personal, so the design had to be calm and exact, with no clutter and no gamified noise.

I built Fraghab for myself, but the wall I hit is a common one. Within months of opening the doors to early users, 415 collectors had signed up to log wearings and build layering combos. The collection manager was becoming a platform.

Fraghab shipped its first roadmap feature at the end of March 2026, and what followed was the most productive stretch of building I’ve ever had. I kept a numbered backlog with strict dependency ordering, and held to one rule: nothing ships unless I’d want to use it that same day.

The recommendation engine grew up alongside the feature work, through thirteen separate upgrades. It went from one hardcoded forecast to live weather for your city, from three crude temperature buckets to a continuous warmth curve, and from five moods and six occasions to nine and twelve. It learned to weigh base notes more heavily than fleeting top notes, to factor projection and longevity by occasion, and to build a preference profile out of your own wear history. The AI layer also became collection-aware, so asking for “something like my Ombré Nomade but fresher” now resolves to the actual bottle on your shelf.

By June, growth made the next step obvious. The recommendation engine, the part of Fraghab that reads the weather, the occasion, and your own wear history to pick your fragrance each day, works harder than anything else in the app. It deserved a foundation built for a much bigger community. I spent weeks measuring exactly where the app slowed down as more collectors signed on, then rebuilt that engine from the ground up and checked its picks against the original down to the last decimal. Nothing about your recommendations changed except how quickly they arrive.

That upgrade is live today. Every suggestion you get from Fraghab comes from the new engine, and the original still stands by as a safety net. There was no downtime and nothing for anyone to notice, which is exactly how a change like this should feel. I also sized it for the community I expect after the apps launch, not the one it serves now.

Fraghab has been live on the web since late April as an alpha. Since the new engine took over, bugs and errors have been minimal and the app has simply been stable, so there was no reason to sit on a long runway. The doors are open, and anyone can sign up today.

In early August 2026, Fraghab ships to the Apple App Store and Google Play. The rebuild happened in this order for a reason. The engine was designed from day one to power more than a website, so the iPhone and Android apps will run on the same core the site runs on and feel just as fast.

This is the payoff phase for the rebuild. The goal is an application that holds its speed as load climbs, so growth stays boring.

From the first commit to the engine rebuild to the apps ahead, the goal hasn’t moved: build the tool a serious collector actually deserves. If that sounds like your shelf, come join us. The door is open.