# AI Didn't Raise the Bar. It Revealed It. #AI *Last Updated: March, 2026* It doesn't matter if Anthropic Cowork or OpenAI Frontier actually work as marketed right away. What matters is simpler and more uncomfortable: the expectations from a knowledge worker have shifted forever. ## The Relay Layer Peter Drucker [wrote in 1988](https://hbr.org/1988/01/the-coming-of-the-new-organization) that whole layers of management "neither make decisions nor lead." They existed as relays, human routers passing faint signals up and down the org chart in a pre-information age. Their value wasn't in what they thought. It was in what they carried. Information technology promised to flatten the relay layer. It did, but only partially. The siloed systems, fragmented data, and the coordination overhead that came with them often backfilled what efficiency removed. The middle didn't disappear; it adapted, finding new work in stitching together what the systems couldn't. AI is different. It doesn't just make carrying cheaper, it makes the stitching itself automatable. > **AI finishes what information technology started.** If your role was to relay, summarize, or repackage, to be the person who gathers inputs from three teams and stitches together a slide deck, the market has already decided that a machine handles that now. Not perfectly. Not yet. But well enough that the expectation has moved. What's left is the part that was always supposed to be yours: the sense-making, the hard calls, the pattern recognition that only comes from years of showing up in your own context. The stuff no one could template for you, even before AI existed. ## The Real Blocker Most people aren't blocked by the technology. They're blocked by themselves. By the inability to articulate what they actually do, why they do it, and what good looks like when it's done. I've been there, not staring at a blank page, but the opposite. Generating dashboards faster than I ever have. Prompt in, glossy chart out. Iterate. Refine. Another visual. Everything looked credible, the insights looked polished, and somewhere in the middle of a third round of charts, I had to catch myself: Who is actually reading this? Why does it matter to them? What do I want them to do next? Those questions had nothing to do with the tool. They were the questions I should have asked first. But the fluency of production let me skip them because the output looked like progress, even when the thinking hadn't started yet. AI doesn't solve that. It exposes it. The professional who hides behind frameworks and borrowed language, who presents someone else's thinking in their own font, gets flattened. Not by AI directly, but by the new baseline. > **When everyone has access to the same summarization engine, the summary is no longer the product. The thinking is.** ## Compliance to Craft This is also a liberation and not in the motivational-poster sense. Compliance was always a diminished form of knowledge work. The processes, frameworks, and templates existed because information was scarce and coordination was expensive. > **Following the playbook was rational when the playbook was the only way to move.** But most professionals in those roles had more to offer than the role ever asked for. The job description just never made room for it. That's what's shifting. When the process-following is automated, what remains isn't a lesser job. It's the job the role was always gesturing toward. The analyst who doesn't just show you the chart but tells you which number is wrong and why. The strategist who walks into the room and names the thing the client couldn't say out loud. The manager who reads what's not in the data. None of that was ever templatable. It was just crowded out by the work that was. This is genuinely harder for people who built careers on relay work, not because they lacked judgment, but because they were rarely given the conditions to develop or demonstrate it. That's a real cost, and glossing over it doesn't serve anyone. But the shift itself, from *did you follow the process?* to *did you see something no one else saw?*, points toward something more honest about what knowledge work was always supposed to be. > **The version that hides behind templates gets replaced. The version that shows up with a point of view gets heard.** ## Show Up Like It The organizations you work for have already updated their expectations. The harder question is whether you've updated your own. Because if you're still measuring yourself by how clean your process was, how many decks you shipped, how closely you followed the framework, you're holding yourself to a standard the market has already retired. The expectations have shifted. Not because AI is perfect, but because the baseline has moved. The question was never whether AI would change what's valued. It was whether you'd develop the capacity of seeing what no prompt can surface. Now you have to show up like it.