I’ve been thinking recently of how the Software Engineering world is changing. I think people don’t want to admit that yes, layoffs will happen. There’s just not enough work to warrant all current engineers working full steam ahead with AI. I built a replica of what took 5 months of 4 engineers over the weekend. There’s all the usual caveats of, the spec was complete already, code was complete, replication != genesis, MVP != production ready. I get it, but even despite that, no single surface of traditional engineering product can survive the fact that productivity is exploding. I’m seeing 15 repos from different AI developers achieving the same effect. 20 people in random threads of X saying, that’s cool, but it seems like a replica of Y. I can’t imagine how much GitHub will have to store in the next couple of years, and it’s just getting started. One example is my fork of clui-cc called clui-cc-codex, that adapts the UI to be provider-agnostic. It took an hour maybe, to get it working well and bug-free, with clear cuts of what features work and what don’t. It’s just a different world. It’s just a matter of time until the software corporate world catches up to what we’re seeing in personal projects.
The only advantage existing software companies have now is scale. Features are no longer a factor in what makes people decide to use one service over another; features are free! I’m seeing people (in my bubble) migrate to open-source projects, local hosting, playing with custom-made tooling for pain points in their lives that are trivial with AI.
I’ll be honest, I’ve been thinking of what differentiates one engineer over another. The traditional measures of an excelling engineer are both being overturned yet kept the same. I think impact is definitely the defining factor now, knowing what is worth chasing and supporting, what moves the needle of people’s lives within your company, what really makes things exciting and makes things move.
I told my manager recently “nothing seems impressive anymore”, which isn’t a reflection of my apathy towards cool things. I believe it’s a signifier of what I think I’m capable of. I think what used to impress me are the same things that impress me now: things that other people do that create impact that I don’t understand and don’t know how to do. With AI, everything is within reach. The commodity is no longer the velocity and know-how of how to do things, but instead the idea and execution of the idea in the way that benefits the most people. Engineers who excelled because they were really good technically are gap-closed by engineers empowered by AI. Everyone is a technical expert now. It’s deciding what projects to lead, what to prioritize, knowing how to leverage AI to the fullest extent.
Used to differentiate
- Being really good technically
- Velocity and know-how
Differentiates now
- Impact — what moves the needle
- Knowing what to prioritize
- Leveraging AI to the fullest extent
Path one
Turns my brain off, uses AI to think for me, presses “Accept Edit.”
Path two
Puts effort in to eliminate ambiguity, fact-checks, asks questions, stays aware.
I sit at my desk and see two paths in front of me. There’s the path that turns my brain off, uses AI to think for me, presses “Accept Edit” from a crud prompt that will eventually get me to where I want to be, while the other path that puts effort in to eliminate ambiguity and think through what the agent is doing, fact-checking, asking questions, being generally aware of what’s going on. AI isn’t an excuse to stop trying. Knowing in what to put your effort in, whether it’s learning low-level code or high level ideas, is what will differentiate one engineer from another.