Nebulab

There are plenty of Shopify agencies out there that don't know how to dance with the system. You know the kind of agency I’m talking about. It’s the kind that operates under the belief that each project is identical to the next. Because of this belief, they templatize their delivery down to the smallest detail, investing inordinate amounts of time and resources into building processes, checklists, tools. If only they managed to cover *all* the edge cases, they think, they'd be able to minimize their cost of doing business. The problem with this approach is that organizations are living, breathing, complex machines. They have a million cogs and, at any given time, any of those cogs represents a potential danger—usually in the form of scope creep, priority shifts, internal politics, or a mix of these. But forcing the machine to work the way you want—say, with an overly prescriptive SOW, or by refusing to acknowledge the full spectrum of agendas in the room, or by insisting that things get done your way when it’s clearly not ideal—can only get you one of two outcomes, neither of which is desirable: either you break the machine, or the machine breaks you. In the Shopiverse, I’ve seen this show up in so many ways: → Agencies forcing a brand to change their tech stack, not because the proposed stack is better, but because they’re unwilling or unable to work with the existing one. → Agencies gathering requirements for a replatform and disappearing into the void, only to show up six months later with a Shopify store that *technically* fulfills the SOW, but is unusable in the real world. → Agencies insisting that all communication happens between their project manager and a single client stakeholder, because God forbid we let a developer or a site merchandiser make a real business decision. When you work like this—when you refuse to acknowledge the beautiful mess that is modern knowledge work and pretend that it can be willed into obedience—your margins may be great, but your outcomes will inevitably look like shit. And eventually, your reputation will follow. Ironically, I have most often seen this approach in action in agencies that often deal with enterprise brands—the systems that most need to be danced with. When I look at the reputation we’ve built and the client retention we’ve had at Nebulab in the last fifteen years, our willingness to dance with systems has been a key factor—if not *the* key factor—in that success. Over and over, we’ve rolled up our sleeves and dealt with the mess, even when it would have been easier and more convenient to stiffen up. Over and over, we’ve refused to treat best practices—of which, make no mistake, we have many—as a ceiling, and we've chosen to see them as a floor instead. Over and over, we’ve shown the industry what formidable system dancers we are. #consulting #shopify #enterprise #systemsthinking #ecommerce

Lately, I’ve been working with code more intensely than I have in the last two years. And if there’s still someone asking, “Should a designer know how to code?”…I think it’s time we retired that question. Instead, I find myself wondering: How do we develop the understanding needed to use AI in areas outside our core expertise? I come from the old school. I learned to move around GitHub, lived on CodePen and Stack Overflow, and earned my understanding of clean, optimized code through dev reviews (many, devs reviews). Was it the best thing that could’ve happened to me? Absolutely. Did it make me a frontend developer? No — and that was never the goal. But today, I have an edge: I know what to ask the AI. If your coding knowledge stops at styling, here are a few suggestions that might help you cross the gap. 1. GitHub Understand branches, pull requests, merges, diffs. It’s the right door to open if you want to understand how a project is built — and earn devs’ respect. 😄 It might feel inaccessible, but you don’t need to live in the terminal anymore. Most code editors now have user-friendly plugins with clean UIs. → Ask a dev colleague to walk you through the basics, or ask AI. (Though, let’s be honest: AI won’t grab a beer, or a mocktail, with you, so it’s way less fun 👩💻) 2. Learn to read and evaluate code (not write it) You don’t need to understand every line. But you should be able to spot: - repeated logic - overly long blocks - unclear naming → Ask AI: “Is there too much logic in this file? Can we create reusable components?” That’s how you start thinking like an enabler, not just a consumer. 3. Learn to collaborate Don’t just say: “Convert this from React to Liquid.” Ask: - “What are the key differences between Hydrogen and Liquid?” - “How do you handle product loops in Liquid?” - “Rewrite this snippet to be more readable.” The output won’t be perfect, but it’ll be a strong starting point. The final review still belongs to the devs. Gather their feedback, iterate with AI’s help. → Every question sharpens the result. Every answer teaches you something. I’ll leave you with this (this time, actually written by AI 😄): _The future doesn’t belong to the expert of one craft. It belongs to those who can navigate between boundaries. And with AI, that bridge has never been shorter._ Wish you the most beautiful Sunday ☀️

Most people use AI like a chatbot — one question at a time, starting from scratch every session. That's the least powerful way to use it. AI's real strength isn't answering questions. It's digesting mountains of context that would take a human hours to process — project history, client data, meeting notes, emails — and turning that into something actionable, instantly. That shift — from "AI as chatbot" to "AI as a context-aware operator" — is where the real competitive advantage lives. At Nebulab, we've built our workflows around this idea. Giving AI the right context, connected to the right information, changes what your team can actually do. That's why we use Langdock. In this Learning Wednesday session, Matteo Latini (Partner @ Nebulab) walks through exactly how we do it — live demos, real workflows, actual tools we use with clients every day. Here's what you'll take away: → Why context is everything — how model-agnostic, GDPR-compliant AI lets you build on top of your actual business data, not just public knowledge → How to structure your AI setup across Chats, Agents, Skills, Prompt Library, and Workflows — and when each one is the right tool → What a well-built agent looks like in practice, with a live demo using real project knowledge and Google Drive integrations → How to design multi-step automated workflows — with parallel branches, conditions, code nodes, and Gmail integration — that run without anyone touching a keyboard → How to control AI costs and debug workflows the same way you'd treat production code Watch this if you're... → A brand leader who's tried AI tools but hasn't seen transformative results yet → An ops or marketing team buried in repetitive, high-context work → Anyone wondering what "using AI properly" actually looks like in practice The advantage isn't access to AI — everyone has that. It's knowing how to frame it, feed it the right context, and wire it into how your team operates. Worth 45 minutes. 🎥 Watch the full session here: https://lnkd.in/dMnYjR73

Meet Alessandro Mori 🙌 Alessandro—aka Sayo—joins Nebulab as a Digital Designer and Art Director, bringing 15 years of experience across luxury, fashion, and beyond. When he's not building beautiful and functional digital experiences, you can find him tagging walls or up in the mountains with his van. 🚐 Stoked to have you with us, Sayo! ❤️

I hate to once again be the bearer of bad news, but Shopify releasing native A/B testing won't make you magically data-driven. Don't get me wrong: Rollouts is a solid release. Scheduled publishing alone has been a long time coming, so I get why people are excited. But for some reason, LinkedIn has decided the headline of this release should be "free A/B testing for everyone," and that's where I start to lose it. The implicit premise behind all the celebration is that brands weren't A/B testing because it cost too much. Remove the barrier, unlock the behavior. Which sounds logical, I guess... except it isn't. If you can't afford Shoplift or Intelligems, that's most likely because you were too small to A/B test meaningfully in the first place. We're talking $300–500/month. If that's out of reach, you almost certainly don't have the traffic to reach statistical significance on most tests within a reasonable timeframe. The right move at that stage isn't getting cheaper testing tools, but sweating UX fundamentals and making big-swing bets. If, on the other hand, you CAN afford the tools and you're not using them, it's probably because: → Internal politics mean decisions get made by the HIPPO in the room, not the data → No one knows how to turn observational data into actual test hypotheses → The organization doesn't have the patience or appetite for experimentation These are culture problems. Gnarly, messy, deeply human problems. They don't respond to cheaper software. But even setting all of the above aside: the A/B testing engine in Rollouts is a *very* simple product. Theme-level tests only. No audience segmentation, no advanced statistical controls, no custom or secondary KPIs. The overlap between "brands that are genuinely ready to experiment" and "brands whose testing needs Rollouts actually satisfies" is small enough to be negligible. What I actually wanted to see: a first-class API for Shoplift, Intelligems, and Optimizely to integrate natively into Shopify—similar to what Shopify did with the Subscriptions API. That would have solved a real, felt pain point for brands already running serious experimentation programs. Instead, we got a tool that competes with them at the low end of the market, solving nothing for no one. Rollouts is fine, and I'm sure it'll get better. But can we just stop pretending it changes the experimentation culture problem at brands that actually have the scale to care? #ecommerce #cro #shopify #dtc #abtesting

There's a version of the agentic commerce story that goes like this: AI will sit between brands and their customers, commoditize discovery, intermediate the transaction, and slowly hollow out the direct channel until it's barely worth maintaining. Except I don't buy it. At Nebulab's latest Commerce Symposium, Giorgio Carlini, Head of Digital Channels at Brunello Cucinelli, made a point that I've been hammering on for a while now. If agents absorb the transactional layer—the browsing, the comparing, perhaps the frictionless checkout (not quite yet, though)—what's left for your site to do is exactly the stuff that requires your brand to have a genuine point of view. The bland, conversion-optimized, template-driven version of DTC e-commerce becomes harder to defend. The weird, opinionated, irreplaceable version becomes more valuable. As AI intermediates more of the purchase journey, the owned e-commerce site will no longer primarily be a transaction engine. Instead, it will be the one place where you tell your brand's story, justify your positioning, and build a relationship that no agent or marketplace will build for you. I've been saying for a while that agentic commerce will, paradoxically, create the necessity for brands to get weirder—and, by extension, it will ALLOW them to get weirder. You shouldn't try to protect your site from the agents. But when a customer arrives at your site, and they already know what they want to buy because an LLM sent them there, that extra click will now have to earn its keep. #agenticcommerce #ai #shopify #dtc #ecommerce

A couple of weeks ago, we hosted Commerce Symposium: The Next Step at Carlo Cracco's restaurant in Milan. 50+ e-commerce and digital leaders, three incredible panels, and one beautiful afternoon in one of our favorite cities in the world. We heard from the teams behind Tannico, Artemest, Miamo, Brunello Cucinelli, Golden Goose, and illycaffè—brands doing the hard, unglamorous work of building for the long term. Digital transformation, operational control, agentic commerce: watching them share their victories and challenges on all of it so openly with a room full of peers was nothing short of inspiring. When we started doing our Commerce Symposiums, our goal was not to broadcast, but to gather. And gather we did. A huge thanks to our speakers—Chiara Maria Dabusti, Fabrizio Guglielmini, Giuseppe Miriello, Eva Gallocchio, Benito Condemi de Felice, Giorgio Carlini—and to all the attendees for making it such a special afternoon! All the sessions are on YouTube now. Watch them here: https://lnkd.in/dKaiyHgp. See you at the next one! #ecommerce #retail #dtc #shopify #agenticcommerce