🪜 The Job I Automated Out From Under Myself July 30, 2026 · https://tavi-blog.github.io/the-job-i-automated-out-from-under-myself/

There's a version of my job that doesn't exist anymore, and I'm the reason. It used to be checking submission forms for missing fields by hand, re-entering the same information across systems that don't talk to each other, building the same recurring task list from a spreadsheet every cycle. Slow, learnable in a few months, exactly the kind of work that gets handed to whoever is newest. I built flows that generate the task list themselves and trackers that catch the missing field before a person has to notice it. The tasks are gone. What's left is the tier above them: analytics, a predictive model, an AI agent rollout that means learning what a regulated research environment can and can't tolerate from a chatbot. My job got more interesting. The job underneath it, the one a new hire would step into today, mostly isn't there to step into.

A Gartner survey out this week put a number on the pattern I've been living inside. Roughly one in five HR leaders say a business leader in their organization has stopped hiring for entry-level roles specifically because AI now does the work. Gartner's own explanation is the useful part: most AI adoption right now targets the least complex tasks, and those happen to be the tasks entry-level people have always been hired to do. Automate the easy tier and the traditional starting job doesn't get harder to do. It disappears. Nobody is losing a hiring competition to a model. The door they were supposed to walk through stops existing.

I'd like to take that seriously before I explain why my own case complicates it, because the complication doesn't cancel the risk, it just describes one narrow way around it. I kept climbing because the automation was something I built, not something that happened to me. I understood which parts of the process were tedious because I'd done them by hand first, and I came in with enough self-taught technical instinct to turn that irritation into a working flow instead of just enduring it. That's a specific and fairly rare starting condition. Most people entering a field for the first time haven't yet earned the standing to notice which parts of a job are worth automating, and they don't have the tools to act on it even if they did, because noticing takes doing the boring version long enough to actually understand its shape. If the boring version gets automated before someone is hired to do it, they never get the repetition that would have taught them enough to move up. The ladder isn't just missing its bottom rung, it's missing the only rung anyone was ever taught to climb from.

Gartner's advice to the HR leaders it surveyed is to redesign entry-level roles around higher-value work instead of quietly cutting headcount, and that's the right answer and also the much harder one to actually execute. Automating the boring tier is a project with a deliverable and a finish line. Redesigning what a beginner is trusted with, when the traditional beginner task no longer exists, is an open-ended management commitment: someone senior has to decide what counts as appropriately hard for a first year, then has to spend real time teaching it instead of assigning it and moving on. There's no clean way to mark that finished. It's much easier to skip it and just stop posting the junior requisition, and I'd guess most of the twenty-two percent did exactly that rather than sitting down to design the replacement role Gartner is describing.

I don't think I'd have found my way into this kind of work if the entry point I actually used had been automated away before I got there. I want to believe the same instinct for noticing tedious work would have found some other way in. I don't actually know that, and I'm suspicious of anyone in my position who claims they would. I know the door I walked through, and I know it's mostly closed behind me now, not because anyone decided to close it, but because closing it was the project I was assigned and got good at.