11 things that helped me in my vibe coding journey
from building slop, to actually good products that can become money-making startups:
- start with questions, not a full PRD in the beginning, PRD should come after AI knows what you want to build
- iteration can't be skipped no matter what, the one shot builds are just a meme, almost always you'll need to iterate for days and weeks
- creating a great ui is a process, lots of variation testing, observing things to the maximum detail, otherwise it will not look that great
- you will need to learn a lot of different things, each project will push you to grow, enjoy the process during the initial learning phases, but then focus on building one main project when you are ready
- using ai just for building is only the tip of the iceberg, use it for distribution, content, marketing, flywheels, business dev, team tools and more
- as you start to master many different branches of ai, from generative, to agentic, to vibe coding, and data related work, you'll have in your arsenal capability to work like you're a one man army
- however, you will quickly realize that switching that much context is not an easy task, and humans are not made for trying to do so much context switching
- you will have to get to a level where you can then build a team of agent operators, a rather small team, if you want to get great stuff done
- this is the new way of building:
doing it solo until you can hire a small team and each person taking still more than one skillset, but not everything
- the repetition definitely counts in ai related skillset too, and the more you spend time with agentic tools and LLMs, the better you'll get at directing stuff
in the meantime models will constantly get better, so mastering these stuff is a very interesting phenomenon because your output quality curve will go up like crazy, due to models getting better too together with your own skillsets
- raising money from VCs is the old way, the new way is building a great product and tokenizing and let the fees accrue to build even more, as you can do this in your own freedom whereas going to VCs and getting an investment is at least 6 months to 1 year of work, which is not suitable for this pace where if you are not fast, someone else will build what you are building
for tokenization not only your product must be great, but you must be great at attention building: so you must learn that as a skill too, and try to master it further through the use of generative ai & unique visions