![[images/Behind The Scenes Of Lovables Insane Design (ft. Lovable Team).webp]] **Creator:** Enrico Tartarotti · **Published:** 2025-05-07 · **Length:** 22:04 · [Watch on YouTube](https://www.youtube.com/watch?v=ZoVLS-Wxr1s) > *In this episode sponsored by Lovable I went to their HQ to go behind the scenes of the design and engineering behind Lovable, the future of coding and vibecoding and what happens now that everyone can build.* (from the video description) A sponsored, documentary-style office visit mixing Enrico's narration with short interviews with Lovable designers and engineers. Most interviewees aren't named on screen in the captions, so quotes below are attributed to "the Lovable team" unless a name is spoken. The auto captions garble several words (Lovable appears as "Love," "Lavable" and "doable," vibe coding as "VIP coding," Supabase as "superbase," Waymo as "Whimo," ChatGPT as "chajbt," and the outro's "I'm Enrico" as "I'm a Rico"); corrected spellings are used below. The figures quoted (revenue, team size, Google's share of AI-written code, WhatsApp's headcount) are as stated in the video. Chapter times come from the video description, with sub-section times from the transcript. ## 1) Detailed outline ### [0:00] Intro - Cold open with clips from the visit: a solo founder who built a product in two months and raised $500,000 while his biggest competitor raised $50 million without launching, and a designer on putting "the work of others in the foreground." - **0.23%** of the world's people know how to code, Enrico says, yet every app, website, video game, device, the financial system "and Minesweeper" were built by them, and seven of the 10 richest people are in that group. - **Vibe coding** removes that privilege: describe an app and watch it get built in minutes. He asks what the world looks like when everyone can build, and what happens to developers, some of whom call it "the vibe coding mind virus." - He flies to **Stockholm** to visit **Lovable**, which he says went from zero to $10 million in revenue in 60 days by letting people build products without coding. ### [1:40] Inside the HQ - **Amir** shows him around; they drink coffee with "performance-enhancing mushrooms." - The office isn't the typical open-plan startup layout. Two rooms are brand new because the company outgrew its space, another was added "yesterday," and the team went to look at a whole building of their own two days earlier. ### [2:39] What are people building? - **[2:32] A short history of talking to computers.** **Assembly** (how Pokémon Ruby and the Space Shuttle software were written) needs a complex block just to add two numbers; **Python** does it in a readable line; now you can skip code entirely. His demo prompt: a calculator app that's a 2D game in the style of **Pokémon Emerald**, where the character collects numbers and operators. - **Fun projects:** an iPad app styled as a vinyl player for Spotify music, built because a similar TikTok app required a subscription (the builder's point: it's free and runs locally); an AI hand-tracking camera demo; and a 3D **Doom** clone with uploaded assets built by **Alex** from the team. - **[4:43] Real companies.** A founder built a clothing search app he calls "**Perplexity for shopping**," which also generates an AI image of you wearing the clothes. He built it in about two months, raised $500,000 and launched, while the biggest competitor had raised about $50 million over a couple of years without launching. - The team says the future is "not build a website but how do you build a business," so Lovable is thinking about distribution, marketing and analytics too. - **[6:12] Limits and internal tools.** You can't prompt "build me GTA 6," but people are shipping real apps, and established companies use vibe coding for **internal tools**. A team member recalls hating an internal tool at **Amazon** that no one had any incentive to improve. ### [7:07] What happens now that everyone can build? - **Breaking the need for scale.** Software is expensive to build and cheap to scale, so a $300,000 app that solves a problem for three people would cost each of them $100,000. Vibe coding starts to break that rule. - Roughly 60% of users build for others and 40% for themselves. One researcher saves about three hours a week with a Lovable app for a specific data-crunching task. - **One-to-one software.** The team expects many people to run custom software built and maintained just for themselves. The best builders for a niche will often be the obsessive person with taste who has the problem, not an engineer, which they predict will cause a "creative explosion." - **Faster to build than to search.** A team member imagines it taking less time to create an app than to find the best one. Since the value of apps like **Uber** or **Spotify** is in the service, people could build their own front end, for example an allergy-first food-ordering app on top of an **Uber Eats** **MCP server**. - **Who builds?** About 80% of Lovable users are non-technical, but developers and experienced builders make the most advanced products, because you still need to understand users, databases, API calls and dependencies in your head. - **[10:57] Floor and ceiling.** Traditional coding has a high floor and an endless ceiling; tools like Lovable have a much lower floor but a ceiling on complexity. Lovable is pushing both, aiming to be able to build 95% of SaaS starters before focusing more on the ceiling, with work on vertical integration and better models. - **Ideas versus execution.** If building gets commoditized, the team says distribution, product taste, knowing what's worth solving and talking to users become what matters. ### [12:50] Behind the scenes of how Lovable is built - **Overheard at lunch:** designers debating "teams" versus "workspaces" as the name for a new feature. - **Hiding the code.** Lovable has increasingly hidden code because it scares non-technical users into thinking they must understand it. Enrico compares it to self-driving cars: some **Waymo** cars still have a steering wheel and pedals, but eventually you remove the old controls to move people to the new world. - **The aha moment.** Because AI output isn't deterministic and still makes errors, the team obsesses over giving a non-technical newcomer a working result, and not an error, on the first prompt. - **Explaining technical pieces.** When Lovable introduces **Supabase** (a back end) or **GitHub**, users may not know what either is, so the design challenge is making highly technical things understandable. - **UI first.** Normal development builds features, architecture and data storage first and the UI last; Lovable deliberately does the UI on the first prompt so the AI doesn't have to make "a gazillion assumptions" about authentication, APIs, requests and storage from a vague prompt. It uses a very opinionated stack, and the exact approach is "the secret sauce." - **Scale.** The hardest technical part is scaling to hundreds of thousands, maybe a million users; about 80,000 projects are built on the platform every day. - **[16:06] Beyond the chatbot.** Probably 80% of AI today is a chatbot because **ChatGPT** made it the default. A designer finds chat limiting for non-visual thinkers and sees potential in mixing text with visuals (sketching an idea), and in losing the big left-hand prompt panel in favor of the AI suggesting two or three website options to pick from. - **[17:11] Sponsor note:** the video is in partnership with Lovable, which offers viewers extra credits (code **ENRICOYT20** in the description). ### [17:26] Is coding dead? - Developers call vibe coding a virus and point out that vibe-coded apps may be insecure or hard to scale, which Enrico says is partly true. - The team's answer: "slope is more important than intercept." Models have gotten much better and cheaper in 24 months, so they bet on the trend toward fuller, more hardcore full-stack software. - **Developer anxiety.** Computer science graduates struggle to find junior jobs and openings are declining. Enrico cites Google saying 80% of its code is now written by AI, and **Mark Zuckerberg** saying Meta would probably have an AI that can effectively be a mid-level engineer in 2025. - Every Lovable person he spoke to was technical. Their views: evolution never stops, so adapt; the best builders will still be engineers for a while; the safe paths are being multifaceted (growth, taste, architecture, distribution) or doing very low-level hardcore engineering. - One team member predicts AI doing all cognitive work, changing the world more than the Industrial Revolution, and moving toward a **universal basic income**-style model. ### [19:34] What happens to companies now - Tiny teams with huge results: **WhatsApp** reaching a billion users with 19 people (as stated), and Lovable going from zero to $20 million in six months with hundreds of thousands of users and fewer than 30 people. - **Everyone is technical.** Almost all of the team is technical even outside engineering roles, which makes communication with teammates and understanding partners easier. - **Drive and culture.** A small team avoids big-team communication costs; Lovable caters lunch and dinner so people stay late and eat together, and there's a running joke about wanting beds in the office (their own hotel). - Enrico expects comments about office beds and mushroom coffee, but argues this attracts people who really want to be there. - **Regular people.** Sitting through the weekly demo, he realizes that behind stories like **Steve Jobs** in the garage or *The Social Network*, these are ordinary people with a mission who work hard and have fun. - **[21:50] Outro.** Points to another video on how the AI wave started on September 30, 2012, and repeats the Lovable credits offer. ## 2) Things mentioned ### AI and building tools - **Lovable** (vibe-coding app builder; Stockholm HQ; sponsor; code ENRICOYT20 for extra credits) - Vibe coding (and the "vibe coding mind virus" criticism) - **Supabase** (back end Lovable connects to) - **GitHub** - **MCP server** (example: a hypothetical **Uber Eats** MCP server for a custom allergy-first ordering app) - **ChatGPT** and the chatbot as the default AI interface - AI models getting better and cheaper ("slope is more important than intercept") - **Flask** (flask.do), Enrico's video collaboration tool ### Programming and computing references - **Assembly** code and **Python** (adding two numbers) - **Pokémon Ruby**, **Pokémon Emerald**, Space Shuttle software - **Minesweeper** - **Doom** (a 3D clone built in Lovable) - **GTA 6** (what you still can't prompt) ### Apps and products built or discussed - iPad vinyl-player app for Spotify music (free alternative to a paid TikTok-promoted app) - AI hand-tracking camera demo - "Perplexity for shopping" clothing search app with AI try-on images ($500,000 raised in about two months) - A researcher's data-crunching app saving about three hours a week - Internal tools at companies, including a disliked internal tool at **Amazon** - **Uber**, **Spotify**, **Uber Eats** (the value is the service, not the app) - **Waymo** self-driving cars (steering wheel and pedals analogy) - **WhatsApp** (a billion users with a tiny team) ### Design and product ideas - Only 0.23% of people can code - Breaking the need for scale; personal one-to-one software - Building an app being faster than searching for one - Low floor versus limited ceiling; aiming for 95% of SaaS starters - Ideas versus execution, distribution and product taste - Hiding code from non-technical users - Getting users to the aha moment without errors - UI first, then functionality; a very opinionated tech stack - Beyond chat: mixing text and visual input, AI suggesting design options - "Teams" versus "workspaces" naming debate - Small, driven, all-technical teams; catered meals and late nights - Universal basic income ### Company figures cited - Lovable: zero to $10 million revenue in 60 days; zero to $20 million in six months; fewer than 30 people; about 80,000 projects a day; 80% non-technical users - Google: 80% of code written by AI (as stated in the video) - Meta: Zuckerberg's prediction of an AI mid-level engineer in 2025 ## 3) Biographies ### Enrico Tartarotti Italian-born YouTuber, designer and entrepreneur whose channel goes behind the scenes of the design, psychology and stories behind tech. He's building **Flask** (flask.do), a video collaboration tool. Here he travels to Stockholm for a sponsored visit to Lovable's office and interviews its team. ### Lovable team The designers and engineers Enrico interviews at Lovable's Stockholm HQ. The captions name only a few in passing: **Amir**, who gives him the office tour, **Alex**, who built the 3D Doom clone, and people introduced in the hallway (the captions render the names as Leon, Lucas, Luka and Eric). One team member previously worked at Amazon. Lovable itself is a Swedish AI app-building startup that lets people create web apps by describing them in plain language. ### Mark Zuckerberg Co-founder and CEO of Meta (formerly Facebook). Enrico quotes his prediction that in 2025 Meta would probably have an AI that can effectively act as a mid-level engineer. ### Steve Jobs Co-founder of Apple (1955–2011), cited for the famous story of starting the company in a garage, as an example of the mythology around startup founders.