Here’s how to work out if your job is safe from AI
Stop asking “which model is best?” and start asking “what type of work do I actually get paid for?”
Because “hard” work comes in three flavours, and AI hits each one differently.
1) Thinking hard This is original problem-solving. Strategy, analysis, making sense of messy inputs, finding a path where there isn’t one. AI is getting scary good here, fast. If your value is mostly “I can think through complex things”, you’re going to feel pressure, because the cost of that capability is dropping and the supply is exploding.
2) Grinding hard This is volume. Hundreds of tickets, thousands of emails, contract reviews, repetitive reporting, migrating systems. AI loves this. Not because it’s clever, but because it doesn’t get bored. If your job is lots of small steps done carefully, assume automation is already on its way.
3) People hard This is judgement under social pressure. Reading the room. Giving tough feedback. Making trade-offs that disappoint someone. Changing minds. Selling an idea internally. Knowing when “yes” really means “no”. AI can imitate language, but it can’t carry responsibility. This is where humans remain expensive and valuable.
So the test is simple:
Look at your week and bucket your work into those three categories. Be brutally honest. If most of your hours sit in thinking hard and grinding hard, you’re in the blast zone. If a meaningful chunk is people hard, you’ve got a moat.
And here’s the twist founders should care about: The people who win won’t be “AI for everything” people. They’ll be the ones who can route tasks properly, choose when to use a model, when to use an agent, and when to handle it themselves because the risk is social, not technical.
Nobody can map this for your role. You have to do it yourself.
So the question to ask isn’t “is my job safe?” It’s: “Which parts of my job are becoming cheap, and which parts become more valuable when everything else is cheap?”