You Know That Person
You know that person. Brilliant, technically. Nobody in the room is smarter. And somehow nobody in the room actually knows what they built, because they've never had to learn how to explain it to anyone outside their own world. They can get it across to other technical people just fine. But hand them a business audience, and that person doesn't need the whole rundown — they need it packaged.
For a long time, that was completely fine. That was the whole deal in tech. You didn't have to explain yourself well. You didn't have to be easy to follow. Somebody else handled turning your brilliance into English for everybody else — your manager, most likely, or maybe a principal engineer. Somebody whose job it was to translate you.
I'm Darian Glausier, and today I want to talk about why that's dying.
Here's something that actually surprised me when I sat with this idea: it's not dying because the industry had some sudden change of heart about "soft skills." It's dying because of AI, which might be the last place anybody expected communication to become the most important skill in the room.
The Old Deal
I want to be upfront about something. Communication is not a personality trait. I hear people talk about it like it's something you're either born with or you're not — you're a people person or you're not. That's not true. It's a skill. A genuinely difficult one. I've worked on mine for years. Years. And I'm proud to say that at this point, people commend me for it — for the communication, for the emotional intelligence. But it took real work to get here. I still catch myself getting it wrong sometimes — saying something that doesn't quite land, having to double back and say it differently. Everybody does, if they're honest.
Here's the thing though: in tech specifically, you didn't need that skill in the same way, or to the same degree. If you wanted to be a career senior engineer, forever, you didn't need the same communication skills as someone heading toward management, or eventually running the whole place.
Software engineers have had this moniker for basically ever — that we're not great at talking to people. There was a time you could spot one of us across a room before we said a word. And look, there are exceptions. There are people who are great at both. But they were few and far between, and that was fine, because the system was built around it being fine. If you were a good person and you did good technical work, you could lean on the people around you — your manager, maybe a staff or principal engineer — to do the translating for you. That was the arrangement.
I don't want to call it gatekept, exactly. Nobody was standing at a door saying you may not pass without communication skills. But you would hit a wall. Faster than you'd expect, and more visibly than you'd like, the higher you tried to go without it.
There's a whole other conversation here about how differently you have to talk to different people — a business audience wants something completely different from you than a room full of engineers does. That's real, and it's not you being fake. That one's big enough to get its own episode, so I'm going to leave it right there for now.
Communication was never required to survive in tech. It was required to be seen.
Why It's Changing Now
Here's the popular narrative right now, the thing you'll hear in every one of these "what AI can't replace" conversations: AI can't take the human stuff. The feelings. The communication. And yeah, sure — partially true.
But I think people are stopping one step too early with that idea. Here's my pushback: the technical understanding doesn't disappear either. It just changes shape. You don't need to hold every syntax rule in your head anymore — fine, that part's real. But you still need enough technical understanding to look at what AI hands you and know whether it's actually good. That skill isn't going away. If anything, it's carrying more weight than it used to, because now you're the one signing off on more output, faster, than you ever were writing it by hand.
I see this with junior folks too. It's not just "does the code work" anymore. It's "did I consider the right things." That's a communication question just as much as a technical one. You have to be able to articulate what you checked, what you didn't, and why.
I don't think this is abstract. Let me give you some actual numbers, because this is happening right now, not in some hypothetical future. There's testing out of Veracode that found something like forty-five percent of AI-generated code samples introduce real, known security vulnerabilities. Forty-five percent. And that number hasn't been improving. There's a project out of Georgia Tech tracking security flaws directly tied to AI coding tools — six of them in January, fifteen in February, thirty-five in March. That's not a plateau. That's a curve. And separately, companies found that developers using AI assistance were committing code three to four times faster than everybody else, but introducing security problems at ten times the rate.
Here's the thing. Those aren't really AI failures. They're validation failures. Somebody didn't ask the right follow-up question. Somebody didn't look at the code. Somebody didn't communicate precisely enough with the tool about what they actually needed. Somebody didn't catch what it handed back before they shipped it.
Look at the AI-generated code before you put up that PR. Every single time. If you don't, that's on you — you're not just wasting your own time, you're wasting your team's.
AI didn't remove the need to understand your systems. It raised the price of not communicating with them clearly. I think about that every time I review something AI hands me now. It's not paranoia. It's just the job — the same way checking a junior's PR was always the job. The tool changed. The responsibility didn't.
The Job All Along
That's the why-now. That's this episode's whole job — just to name the shift and tell you it's real.
The rest of this series is going to be the how. How you actually talk to different audiences without feeling like you're performing a different person for each one. How you actually talk to AI, because — spoiler — it is not the same skill as talking to a person, even though it feels like it should be. And what all of this means as you're trying to move up, and figure out where you actually stand in it.
But here's what I want to leave you with today. The people who were already good at explaining things clearly — the ones who could say the same thing five different ways depending on who was in front of them — they were never doing some soft, lesser version of the job. A lot of the time, they were doing it better than the people who got promoted past them for being technical alone.
Good engineers with good communication skills have been sitting right there the whole time. We just overlooked them in favor of pure technical skill for a long time. I don't think we get to keep doing that anymore.
I think the industry is just now catching up to the fact that it was the job all along.
I'm Darian Glausier. This is Head First.