Off the Record with De Soi: How a Non-Alc Brand Grew Up

Most of what an operator learns while scaling a brand never leaves the room it happened in: the fires to put out, the buybacks, the production run that goes sideways. That's why we started Off the Record, a fireside series where operators talk honestly about the work behind the curtain.

For our LA session, Sebastian Debrouwere, VP Business Development at DOSS, sat down with Antonio Landa, Senior Operations Manager at De Soi. What followed was an unvarnished walk through how a consumer brand actually rises to the occasion when they get ‘found’, moving from spreadsheets and prayers to a real operating platform. The full conversation runs about 70 minutes and is worth watching end to end.

If you're short on time, here’s the overview.

De Soi is a non-alcoholic aperitif brand co-founded by Katy Perry and Morgan McLaughlin, built on sparkling drinks made with adaptogens and botanicals. Its ethos is "pleasure with restraint." When it launched, the non-alc aisle was mostly beer replacements and sugary mocktails on one end and melatonin and CBD drinks on the other. De Soi set out to fill the space between: a drink that gives you a light lift without the jitters, the sugar crash, or the hangover.

Antonio runs the operations that make that promise hold, and his path wasn't linear. He studied biology, ran new-store operations for a smoothie company, then spent years implementing ERP systems for private-equity healthcare rollups before joining De Soi in early 2024. In that stretch the brand grew to nine flavors and roughly 45 SKUs across D2C, Amazon, and national retail. What follows is the journey underneath that growth.

Every young brand starts the same way. In Antonio's words, it's "spreadsheets, duct tape and login portals," with production in one sheet, inventory in another, and finished goods in a 3PL portal. The team spent so much time on manual processes that, as he put it in our case study, they "couldn't make strategic decisions in real time."

What set De Soi apart wasn't better tools. It was a habit: obsessive structure and consistent nomenclature, a master item sheet where every column meant something. That discipline looked like overkill early on. It turned out to be the foundation the whole business would later run on, and the reason a two-person team could eventually lean on systems and AI at all.

Growth is where good instincts stop being enough. Asked how he solved early problems, Antonio was blunt: "a lot of instinct and a lot of prayers and a lot of educated guesses." The story that makes it concrete is Target. De Soi landed an end-cap promo and went nationwide, celebrated, then hit the fine print: any inventory that didn't sell, they had to buy back. It was 2024, non-alc buyer confidence wasn't there yet, and it didn't move. "Where is all this expired inventory coming from? What do we do with this?"

What mattered was what he did next. The team recovered through distributors like KeHE, rebuilt the footprint "market by market, store by store," and went back into Target the right way, from Très Rosé to the Variety six-pack. Underneath the comeback was a baseline Antonio built by hand: what the business should produce even if nothing changed, with promos and new doors layered on top. Scaling into retail doesn't bury the operators who never slip. It buries the ones who don't turn the first mistake into a system before the next one hits.

For a while, discipline and hustle covered the gap. Then they didn't, and the reason was people, not software. De Soi had turnover, and Antonio felt every departure: "You've invested so much time, knowledge into this one person, and they end up leaving." Losing a colleague wasn't just an open seat, it was institutional memory walking out the door. It reframed the problem into a question: "Can I create a system that knows this, can keep this, and there's no risk of them disappearing on me?" That collided with SKU expansion and omni-channel growth on a two-person team. "It's not working anymore. We're dropping the ball here." A little over a year ago, he found DOSS.

He was specific about why most systems were wrong. Legacy ERPs make you conform to what they've already built, and most only show you what already happened. In our case study he put it plainly: "A lot of other ERP systems were very rigid and you had to conform around what they'd already built. DOSS was pretty much the opposite. It was very flexible and it molded to our business processes." He chose DOSS because nothing else could build a full production module that flowed into the rest of the business. It unified procurement, inventory, and order management on top of the tools De Soi already ran, and pulled the maze of spreadsheets and login portals into one source of truth: a modern operations platform the business runs on, not another system to wrestle.

With a foundation in place, the question changed from "how do I survive" to "how much can two people run?" The answer is a lean team amplified by AI, but only where Antonio trusts it. "I've been through 17 production runs, and every time I learn something new. AI is not going to know what to do in that instance." Where AI does earn its keep is the repeatable, structured work, and that only became possible because of the discipline from stage one. "It was easy enough to upload a master item sheet into DOSS when we did our implementation," and just as easy to feed that clean data into AI. His lesson: "the hardest thing is the data cleanup. It's the thing that takes the most time." That groundwork pays off in the number that matters most. In DOSS, De Soi's 50+ raw-ingredient costs now roll automatically into month-end COGS, giving Antonio live cost and margin visibility at the ingredient and finished-goods level instead of reconstructing it after the books close. This allows them to improve contribution margin with each production run.

That mix of clean data and human judgment is why he adopts AI on the boring work and verifies it on the expensive work. It's the same tension he named in our case study, and where he sees things heading: predictive AI, then autonomy within guardrails, so operators get their time back for the calls that need judgment. That's not a someday idea. It's the whole point of an AI-native platform: instead of a monthly report on what already happened, DOSS gives a lean team a live read on what's coming, and the automation to act on it.

The hurdle is trust, because the technology still hallucinates and no one wants to hand over the keys without knowing it will hold. For De Soi, getting across that gap comes down to the same discipline that got them here: keep the data clean, keep a human on the high-stakes calls, and move the line as the tools earn it.

Asked what he'd tell himself when he started, Antonio didn't reach for a growth hack: "Not everything's going to go right all the time. You're going to mess up. The one thing to do is admit you're wrong, learn from the mistake, and don't make it again." That's the whole point. De Soi grew up by keeping its data clean when that felt like overkill, turning its worst retail week into a baseline, putting a real system underneath the business when two people couldn't hold it, and being honest about which decisions it isn't ready to hand off yet. And all this led them from the lessons of that first end-cap promo to rising to the occasion with a supply chain that delivers their product experience to 7,000 stores nationwide .

Watch the full conversation with Antonio here , or for the short version of how De Soi runs on DOSS, read the De Soi case study .