I have 18+ years of experience as software developer. But I have no experience with Kotlin. And to learn Kotlin, to learn the Android libraries, it is not an easy job. I may need at year of active learning and trying things, before having the confidence to start doing something.
So, I asked myself, how far can I go with AI tools? And I went far!
I created https://howareu.app/
The boring statistics: 126,000 lines of Kotlin across 398 files 45,000 lines of tests across 130 files 3000+ unit tests 50+ production lessons captured 4 months from zero Kotlin experience to production app; 0 lines of Kotlin written manually
Below are the main things that I did and helped me do the product!
Good framework Vibecoding something that big is not an option. You can't write anything in one session. You need a way to keep the context somehow. You need specification. To maintain my specifications I use BMAD framework. It helped me a lot to clear the requirements, prepare the architecture, UI, find any gaps. I used it also a lot for brainstorming session and for marketing. For the latter, it failed
Follow the project rules I use Claude Code with Opus 4.6 (yes, the old version) and keep all my important rules saved in CLAUDE.md file. Every rule in that file exists because I violated it once and something broke. The file grows with the project. This is the key insight: CLAUDE.md turns one-time lessons into permanent constraints. The AI never forgets a rule I put there. I forget constantly.
Keep your documentation up to date I start my session with a custom command that loads all important documents. Enough context, so that when I start prompting, the AI has the basis to make best decisions possible. I end my session with custom command to save all learned, and architectural updates. This way each session is built based on the knowledge from the previous. I have a history, I have an AI that gets smarter with each next session.
Be as descriptive as possible
When you prompt, be as descriptive as possible. I use BMAD to create technical specifications out of my prompts. If the idea is not clear in my head, I do brainstorming session. All possible to minimize the guessing and wrong interpretations during the development phases.
Do you all do code reviews? I do. Usually, with a different LLM. This gives a 'different point of view'. Improves greatly the code quality!
What this is not It is a no-code, but If you know the language, it is worth checking and correcting. With time, the needed small fixes will become less. It is always good to understand and to make the design decisions yourself. It is not effortless. The workflow took months to build. The documentation is extensive. It is not magic. The AI makes mistakes. The difference is that mistakes are caught by the process (tests, reviews, rules, audits) instead of by users;
The takeaway
AI coding tools are not magic code generators. They are force multipliers for engineering process. If your process is "open chat, type prompt, hope for the best," you will be disappointed.
If your process is "document the architecture, define the rules, automate the lifecycle, capture every lesson, review everything critically", the AI becomes unreasonably effective. The investment is not in better prompts. It is in better engineering.
Honesty first
Why am I writing all this?! Because I am desperate! I made this app to help people. Make something good. I want to be sure that I cover as many cases as possible, and that it will really help people. The moment I share in the proper channels that such app exists, I get treated as an intruder, who sells evil!
If somebody would like to help, please, install it, and give me a honest feedback! Thank you!
No comments yet.
