Are We Finally Done Saying AI Will Kill Software Development?
Published August 17, 2026I just got back from Render Atlanta and have been decompressing from all the networking, happy hours, and events. I've been thinking a lot about how this year felt quite a bit different from previous years, especially last year, and especially when it comes to conversations around AI and software development.
Last year, I felt a sort of nervous energy around agentic coding. There were some talks completely focused on it, but not a majority of them. I talked to more people in the hallways who were tired of hearing about AI than who were excited about it. Those who did use coding agents were a little shy about that fact. Not everyone was quite ready to admit whether or how much they used tools like Claude Code or Codex (both of which were still relatively new).
Over the course of the next year, there was a shift though, and I started hearing more and more about developers embracing coding agents and finding a lot of value. I was skeptical back at the end of 2024, but I started finding value too at the beginning of this year.
Then I noticed another shift. The loudest proponents of AI started saying things like "Software development is dead" or "AI is eliminating software developers." Some framed this as a good thing, others as doom and gloom.
I think a lot of people believed them. Online discourse shifted heavily toward discussion of whether anyone needed to bother knowing how to code at all. "Vibe-coding" became either the hottest topic at conferences, or a dirty word developers scoffed at. So-called "thought leaders" online continued to push this narrative of coding agents getting so good, you don't need to know anything about software, and those who do very soon wouldn't have value at all.
This year at Render felt very different. Nearly every talk was about AI in some way. That was certainly a shift. But what was more noticeable was the shift in tone. Talks and conversations focused on AI as a tool, not as a replacement. The physical act of coding may be cheaper now, and may even be disappearing, but we've moved on to understanding that coding isn't everything in software development.
I found that I wasn't so fatigued hearing about AI in every talk on every track of the conference.
I had great conversations around how to leverage AI for mundane tasks (yes, sometimes writing code is just so boring). We talked about how developers can now focus on architecture, and implementation of critical or complex patterns (the fun parts) rather than spending all day typing. Nobody was arguing that software developers are or will be obsolete. I don't think there was even much talk of "vibe-coding" at all this year. Instead, everyone seemed excited to learn about how to use AI in a human-centric way, where we still rely on our hard-earned software skills, and how to avoid the traps of offloading thinking to the machines.
What does this mean long-term? I'm still not sure. Frontier coding models have come a long ways very quickly, and there is still far more uncertainty in the future. However, it was encouraging this year to hear newcomers, juniors, and seniors alike embracing the idea that generating slop isn't the future, and that the future relies on thoughtful, intelligent humans with real skills, using powerful tools to create and build. That sounds to me a lot like what we've always done.
