In this episode, we interviewed Jorge Medina - founder, designer and product builder who runs Groou Studio, a branding studio for funded companies, and is building ramie, a platform where a brand lives as a system of rules rather than a folder of files

Jorge's career runs through paid advertising, creative agencies, product development and creative analytics - always circling the same problem: brands that fall apart in execution

His core argument: AI took over execution, but execution was always the smallest part of the job - the thinking before the doing is the product, and outsourcing the deciding leaves nothing worth executing

Key takeaways:

🔗 Connect with Jorge

I'm Jorge Medina: founder, designer, and product builder. I design brands and design systems for founders who take their brand personally, and I'm building ramie, a single source of truth that keeps a brand consistent everywhere it shows up, from marketing collateral to product screens.

I've spent my whole career watching the same problem from different seats. Paid advertising, creative agencies, product development for branded content, creative analytics, and now my own branding studio. Everywhere I went, the problem nobody solved was the same: keeping a brand consistent at scale, across everything it touches. I even took a first swing at it with easybranding, a design startup I ran for a year and a half before killing it. Wrong business partner, right problem. For companies without design talent in-house, brand drift is just a matter of time. The tools don't help non-designers execute with precision, and AI-generated assets are educated guesses, no matter how detailed the prompt or how thorough the design.md file. A brand is precisely the thing you can't keep re-guessing.

Design systems have kept software products coherent for decades: tokens, components, shared rules. But the practice never really left tech, so most businesses still get their brand delivered as static files and a 40-page PDF nobody reads, and things start falling apart the moment the designer leaves the room. I apply design system thinking to every single thing I do, pushing brand systems past UI into editorial territory: how the brand writes, builds a deck, art-directs photography. Roundr, a client of two years, runs on this. The identity was designed upstream, and messaging, marketing assets, and the website all derive from one system, so the next asset never depends on the design skill of whoever is creating it. Touchpoint fifteen looks like it came from the same mind as touchpoint one. The tooling for this barely exists, which is why I'm building ramie.

AI helps with execution, correct. But execution was always the smallest part of the process, and the most annoying, which is exactly why everyone's so happy to hand it over. Whether you're creating a brand, a marketing campaign, or a product feature, the core issue was never executing the thing. It was knowing what to execute, why, and how. The thinking before the doing. That's what too many senior leaders are externalizing right now, flattered by enthusiastic AI responses that make everyone feel like a genius outside of their domain of expertise, with nobody interrogating whether the thing they just agreed to build should exist at all. In branding this gets expensive fast, because the thinking is the product. A brand is a set of decisions about who you are. Outsource the deciding to the machine and there's nothing left worth executing.

Most of my AI use lives in Notion. My whole business runs there, meetings, projects, tasks, documents, client work, and because all that context sits in one place, Notion AI has absorbed the least interesting parts of building a company: the admin, the summaries, the chasing. That's the workflow I rely on every single week. When I design identities, I barely use AI at all. It's mostly Figma, with the occasional Midjourney session to curate an illustration style, or Figma Weave for photography when a client can't budget a real shoot. Building ramie is the other extreme. I keep milestones and tickets in a Notion database, write the product requirements, and Claude Code picks them up through the MCP connection. I manage it the way I managed my devs when I led product at startups: we work the roadmap together, every day.

Split down the middle. Studio days are client work: messaging documents, brand strategy, positioning, visual systems, website builds, the occasional deployment that eats a whole evening. The rest is ramie, which right now means the unglamorous middle of building software. Sometimes it's grueling, maddening work, but the feeling of making something from scratch is hard to beat. It's a lot of writing and sharpening product requirements, then working the tickets with Claude. I throw away more than I keep, and the key is to always be skeptical of the easy answer.

Visual AI slop gets so overwhelming that demand for real design expertise spikes. AI's literal job is to land on the likeliest average of whatever you ask it for, and the likeliest average is the exact opposite of what a brand is for. Being distinct for the right reasons, in the right way, is strategic and nuanced work. No prompt gets you there. There's plenty AI can do for design, but I don't think generating screens and assets is it. The designers I rate are building custom tools for their clients instead. I did it myself with prcdrl, a procedural texture and motion tool, and that experience is how I realized I could build ramie. The designers who learn to build this way are the ones who'll be in high demand in twelve months' time. They already are, to be honest.

LinkedIn. I've been heads-down building for months and my feed shows it. But I'm going back to building both the studio and ramie in public, and hopefully adding one more human voice to a feed drowning in AI slop. If your brand is starting to matter more than your capacity to keep it true, my studio lives at groou.xyz, and that's a conversation I'll always take. If the drift problem sounds familiar, the ramie waitlist is at ramie.io. Come say hi.