I’m in the middle of learning and building agentic AI systems. The goal: build a multi-agent workflow of customer service agents. While the scope sounds narrow, in practice, it can be right sized into many use cases since many teams are concerned about the handoffs, the reasoning, and quality of output that this technology can provide.
While this work is not leading me to believe AI is going to take everyone’s job tomorrow, it did illuminate for me that the people who aren’t paying attention right now are going to have a very hard time in the future when this is part of our daily work in a much bigger way.
There is a counterargument I keep seeing on LinkedIn: Perplexity’s CEO predicted recruiters would be replaced by AI in six months. Since they weren’t, many believe the entire premise is smoke. I understand the appeal of that logic. It was a poor proclamation and aged like an overly ripe banana. That said, there is a distinct difference between a broken timeline and a broken premise. A lot of people are conflating this in ways that may come at a later cost to them.
CEOs have always been aspirational—and sometimes delusional—about timelines and what the future holds. But a big part of their roles is to help contextualize a vision they are trying to sell. Steve Ballmer famously declared there was no chance the iPhone would get significant market share. That one was
just wrong. In 1999, Webvan was launched, and only lasted for 20 months after around $800mm in total funding. While the consumer wasn’t ready for grocery delivery at the time, Louis Borders—yes, the bookstore guy—wasn’t entirely wrong; his timing was just wrong. Webvan effectively built the infrastructure for what is now Instacart, Amazon delivery, and the like.
What should not be in question is the direction. Too much has been built, too much has been proven, and the models are improving at a pace that is observable and measurable. Six months ago, what I’m building now would have been meaningfully harder to do. The gap is closing faster than most people are treating it.
Here is what the LinkedIn observation on the recruiting elimination narrative is missing: Blockbuster wasn’t wrong that people wanted to watch movies at home, but they took their eye off the ball and didn’t evolve that thesis. By the time they figured it out, they were in nostalgia territory.
The people concluding from the missed timelines that the AI narrative is fundamentally broken and not upskilling on AI are making a Blockbuster-level error but for their future careers.
Do I think we’re in an AI hype cycle now? Yes. Do I think a meaningful number of AI-native companies are going to disappear? Yes. This happens with every significant technology shift. A bubble and a shift are not the same thing. The internet bubble burst spectacularly in 2000, and it was painful for us who are still brushing off the skid marks from that one. That said, the internet did not burst with it.
What I have experienced over the past weeks is not a gimmick, rather the build of what is to be an effective and efficient multi-agent system that reasons through a problem, hands off context, and produces quality outcomes in a way that six months ago would have been meaningfully worse. The models are improving quickly.
The question is not if technology will change how knowledge work gets done, rather whether you will be someone who shaped how it is applied or are too late to the party.
Take recruiting, since that’s where the “redundant in six months” prediction landed. That didn’t happen. I’d argue it didn’t happen not because AI isn’t capable, but because the prediction fundamentally misunderstood what the job of a recruiter is. Good news, it is something I happen to know very well.
I don’t want to negotiate a job offer with an AI bot. I don’t think most candidates do either. The moment in a search where trust gets built and where a candidate decides if the company represented is worth the risk of leaving somewhere safe lies with a human connection and that won’t likely change in my lifetime.
AI can do the mechanical portions of recruiting— driving market research; surfacing the “purple squirrels”; building a search strategy; tracking pipeline; synthesizing patterns across hundreds of data points—faster and at a higher volume than any individual recruiter. The recruiters who will thrive in three years are the ones who figured this out early, embraced AI, and leveraged it to reallocate their time accordingly so they can focus on real, human connection.
That’s not only a recruiting story, rather a story for every knowledge worker in every function who is trying to figure out what this moment requires of them.
There are two skills that are going to matter, and they are not in competition with each other:
1. AI fluency beyond prompt-and-paste. This will meanunderstanding how to build workflows, how to architect systems, how to evaluate output critically. The people who will be valuable are not the ones who can only use the chat box. They are the ones who understand what the technology can do structurally and can apply that to real problems.
2. What AI genuinely cannot replicate: EQ. This reads as the ability to read a room, understanding organizational dynamics, knowing when the data is telling one story while the humans in front of you are telling another. It is about empathy, solid judgment, and context that lies in real-life relationships. AI
can’t touch that.
The knowledge workers who will struggle are the ones who never got fluent in the
tools and never doubled down on what AI can't replicate.
The Perplexity CEO was wildly wrong about six months. And his timing will likely still be wrong in another six months. I don’t know the timeline, but I do know this: I built something last week that works. While it is not a finished product and, at times, made me want to toss my computer out the window, it works at a level that would have taken a significantly larger team significantly more time not that long ago.
The clock on AI may be wrong, but the direction likely isn’t. The question is what you’re going to do with the time you still have to get ready.