Vibe coding has produced two loud and equally unhelpful camps. One says anybody can build software now and the profession is finished. The other says the output is garbage and none of it matters. Having spent time with these tools, the truth is duller and more useful.
They collapse the distance between an idea and something clickable, which is genuinely valuable, and they do very little about the part where software has to keep working. Lovable is currently among the best of them, and its official learning material reflects both halves of that. Start with what is good. The documentation is properly done.
Quick Start gets you to a published app in about ten minutes, and that first moment where a non developer watches their own idea appear in a browser is the entire reason this category exists. Idea to Product walks the fuller arc from concept to live application. There are sections on features, integrations, prompt engineering and a glossary. The prompt engineering material is more considered than the name suggests, dealing with how to describe what you want in a way the system can act on, which is a real skill and one that people flail at without guidance.
The integration coverage is the other strength, with Supabase, Stripe, OpenAI, GitHub, Clerk, Resend, n8n and a long tail of others documented, because the difference between a demo and a product is almost always auth, payments and a database. The video library is where it gets loose. It is a tagged collection rather than a curriculum. There are quick guides, multi part builds, live streams and beginner walkthroughs, and there is emerging material that looks like a structured course, but as it stands a beginner arriving at the videos page has no obvious first step and no sense of what order anything goes in.
Compared with the documentation, which clearly had someone thinking about sequence, the videos feel like a marketing asset that accumulated. The live streams are the best of them, because watching somebody hit a problem and work through it in real time teaches more than a rehearsed tutorial ever does. Then the money, which is the part the learning material handles least honestly. The free tier gives you five daily build credits up to thirty a month and twenty cloud credits.
That is enough to try things. It is not enough to build anything. Pro is $25 a month for a hundred credits, and a genuine first project routinely consumes eighty to a hundred and fifty credits in the first month, with additional credits at roughly twenty cents each. Simple changes cost a fraction of a credit, more complex work costs more, and iteration is where it accumulates, because you rarely get what you want on the first prompt.
Anybody treating this as free learning is going to be surprised, and the tutorials do not prepare you for that. The deeper limitation is transferability. What you learn here is how to work with Lovable. The prompting skill carries over somewhat, and understanding what a database or an auth flow is for carries over completely, but the practical knowledge is platform shaped.
If Lovable changes its pricing or its behaviour, and this category changes constantly, a lot of what you learned becomes about a product rather than about building software. That is fine as long as you know it. The omission I care about most is everything after launch. There is essentially nothing on security, on what it means to hold user data in a database you did not design, on cost control when an app gets traffic, or on how to reason about code you did not write and cannot fully read.
People are shipping real applications with real users out of these tools right now, and the training treats deployment as the finish line. It is not. My 3.6 is for excellent documentation, a genuinely impressive product, and a first hour experience that will make a lot of people believe they can build things, held back by an unstructured video library, credit costs that the free framing obscures, and total silence on what happens once something you generated is actually being used.