you can’t automate this, bro’
What’s the hardest problem in software development? Is it the coding? Is it the tools and the frameworks? Well, let’s assume you are a genius coder… You know all the programming languages in the…
Nov 11, 2025 · 3 MIN READ
What’s the hardest problem in software development?
Is it the coding?
Is it the tools and the frameworks?
Well, let’s assume you are a genius coder… You know all the programming languages in the world. All the design patterns and the tools.
Knowing all that… Someone asks you to build an app.
Is a tool for insurance companies.
It includes 250+ business rules, 35+ edge cases and 20 different countries.
(Here are only a few of them: risk scoring, premium pricing, coverage eligibility, claim validation, policy renewals, fraud detection, canceled policies, different minimum coverage, driving regulations, healthcare systems, or claim caps per region… etc)
With all your knowledge… Would you be able to build this app in a few hours? Or days?
Probably not!
That’s because the biggest challenge in software engineering is not downstream (languages, frameworks, tools, etc).
It is upstream!
It is understanding what to build exactly, dealing with tradeoffs and ambiguity… And putting all that together in a working application that doesn’t go down overnight!
AI is busy trying to automate coding… But all it automates is boilerplate creation.
LLMs can hide part of the complexity of the apps we are building behind fancy prompts… But they can’t eliminate it… Let alone actually deal with it!
Which is exactly why AI won’t make developers obsolete anytime soon.
Low code and no-code promised to do just that decades ago… The first low code tools were created in the 1980s (Oracle Forms, Microsoft Access, Lotus Notes).
35 years later, there are more developers on planet Earth than we ever had before.
Fred Brooks talked about this in The Mythical Man-Month (1975) and later in No Silver Bullet (1986).
He mentions two kinds of complexity in software:
Essential complexity → the inherent difficulty of the problem itself (e.g., business rules, domain logic).
Accidental complexity → the extra difficulty introduced by our tools, frameworks, or architecture choices — not by the problem domain itself.
Even if AI fulfils its promise and automates ALL the accidental complexity out there…
You need developers to deal with the essential complexity of the problem!
Okay, this is argument 157 I am making in favor of software devs. Hope is convincing enough for you to pay less attention to “vibe coding” gurus and more attention to software development fundamentals (which are here to stay).
Anyway…
November spots are almost full. Book a call to see if you are a fit here.
Btw, Bogdan and I are thinking to put a cap on the students we get on until the end of the year.
About 7 to 8 spots max.
After that limit is reached… We will close admissions until January 2026.
If you want us to help you level up, better act sooner than later. Book a call here and let’s chat :)
I know people who every year come to me telling me they've been watching us for years, and they want to sign up.
But when it comes to taking action… They bail out.
Meanwhile, developers who joined 2 years ago are already at technical lead level and asking us to put together a program for them to get to architect/staff engineer.
The difference is in what you do… Not in what you say you will do!
Either way, I wish you all a very productive week!
Best,
Dragos
P.S. Ohh btw, today is my and Bogdan’s birthday. Expect a big party and pictures coming up next week!
