2026: Coding is dead; Learn to let it go
How to stay employed in 2026
Software engineers are more wizards than programmers these days. Without argument, the current state-of-the-art models from Anthropic and OpenAI are magic. Given a task, they will outclass the vast majority of developers and are fundamentally rewriting what being a software engineer actually means. One thing I’m sure of: writing code is obsolete. So stop writing code. Learn to let it go.
I’ve had an insecurity working in an industry full of very smart people who are incredible programmers. I’m a good engineer, but I’m no 10x dev. What I am good at and love doing is problem solving and designing product. Adopting the edge of AI tooling has been an absolute blast. My mindset and my output has been completely changed over the last couple years of building startups. I’ve progressively written less code and have shipped more than ever. This is not vibe coding, this is Product Engineering.
The timeline has looked like: GPT 3 and Copilot giving us line completion. Sonnet 3.5 and Cursor that gave us file completion. More recently Opus 4.5 and Claude Code that gave us feature completion. The trend is one way. Better models, more autonomy. Just look towards Gas Town.
Even if models never improve beyond the current generation, the threshold of competence has already been crossed and we will be mining productivity gains for years to come.
Code is everywhere, and only becoming more prevalent, so how can it be dead? Pre LLMs, writing code was inherently effortful and a byproduct of labor. Highly valuable, hand crafted, laborious. LLMs are the programers printing press, and turning code into a commodity.
Code is now a commodity like oil or steel. Steel is 0.50$ a pound but your smartphone that has a couple grams of steel is worth 1000$. Steel’s price reflects a commodity. A phone’s price reflects utility. So if engineers want to stay valuable, they need to move up the value chain.
My vibe coded app no work
LLMs are amplifiers of skill and knowledge. Give an LLM to a seasoned engineer and they become spell casters. Give it to a muggle and run the other way. The knowledge base of software engineering has never been more powerful, but you still need to know your book of spells. So even the coding is dead, it’s never been more important to know how to code.
The vast majority of SWEs employed were just working on some sort of CRUD app. Codifying business directive into computer operation. We maintained our employment in the complication of computers. We wrote code to solve problems, tediously breaking them down into steps a computer could follow. LLMs have completely annexed software engineers in that process.
Under all that noise, the most employable skill remains the ability to problem-solve, which engineers tend to be quite good at. The challenge for many is, the problem-solving we’re moving toward resembles product management, which engineer-y engineers struggle with. Highly collaborative, soft-skill intensive, and usually operating at a high level of abstraction. In short: we need to adapt into Product Engineers.
In practical terms, there are two paths that have opened up in front of us.
Option 1: Build quicker
Every engineer has become the equivalent of a cracked full stack team. You need to rethink how you ship product to customers. Remove all the layers between engineer and customer. Then ship in days not months.
Option 2: Build complexity
The bar for truly ground breaking work increased an order of magnitude with LLMs. When everyone can up skill themselves to 1000x developers, what constitutes incredible work fundamentally changes. Start thinking bigger.
Learn to let go
The writing is on the wall, coding is no longer where our value is as devs. In a 3 year time span our entire industry has been disrupted, and the trend is not slowing. To remain relevant, we need to move up the abstraction layer. That means being comfortable letting go of writing code. Move faster. Think bigger.
If this resonated with you as an engineer I’m hiring → Product Engineer @ RetailPath