I like to think of the the recent improvements of agentic capabilities as a new colleague that is learning alongside you.

Consider this a surrender, but with the amount of talent, funding and attention poured into them, it is likely they are simply the better colleague than you and will go further than you in the long run.

We’re not here to beat out the AI, we’re here to stay employed long enough that by the time AI takes over the world, we’d be like those baby boomers who have already made their riches and can retire. (of course there is a case to be had that money may become worthless when AI takes over, but premature optimisation is the root of all evil)

For us mostly in junior roles now, just be the “work smart” colleague who knows how to do it, but finds a way to hand it off to a colleague and claim credit.

It is still paramount that you know how to do the work. As mentioned above, to claim credit, you would want to be able to read, understand and explain their work, and tailor it to a way that makes sense to you. A simple way to test that last condition is, can you defend this work when you present it to your higher up, and your higher up questions you about it? All this with unshakable confidence and clarity of thought?

Additionally, you cannot expect to do well in senior role where you manage juniors if you don’t know the junior work: what they do, what they will struggle with, what bad decisions they will make, how to solve their issues, etc. I don’t think that junior roles will be erradicated that soon, but even if they do, and your new “senior role” job is to manage a bunch of “junior” claude instances, it’s the same shit. You’re not going to be good at your job.

So if you’re at work right now, please put in the time to knowing how to do the work, don’t just claude everything, because you’ll find yourself unable to promote. At a senior level, much of the work is decision making rather than grunt work. And that decision making comes from understanding the tradeoffs in the grunt work. The earlier learnings compound.

Unfortunately, in my personal experience that is the software engineering industry, this shit is so saturated that the top roles have, on top of being able to do the grunt work, a requirement of being able to reason about/make those tradeoff decisions as well. But that’s too bad, I love the work, I’ve chosen it, and so all that’s left is to put in the extra work to get there. Don’t stop learning and getting good at what you do :)