Companies may be about to fire exactly the wrong people
Companies may be about to fire exactly the wrong people.
Not because AI can’t do a lot of their work. It can. But we’re still mostly looking at what people do, rather than what they know.
Think about the person who knows that this particular customer needs to be handled differently. That when X happens, you shouldn’t follow the official process. Or that a number can look perfectly normal in the system but usually means something has gone wrong.
None of that is in their job description. It may not be in Salesforce, the ERP, or the company SOPs either. It’s just something they learned after doing the job for five years.
And this is where I think AI gets really interesting.
For a long time, knowing what should be done and knowing how to do it were bundled together. A great QA person might understand the customer, product and edge cases better than almost anyone, but still need an engineer to actually change the product.
Now they increasingly don’t. The same is true for founders, operators, analysts and domain experts who always knew what they wanted but needed someone else to make it happen.
AI gives their knowledge an execution layer.
That’s a huge opportunity, but there’s also a nasty failure mode: a company maps its processes, automates everything it can see, cuts the people who appear redundant, and becomes more “efficient.”
Except the process map was never the business.
The business also lived in all those exceptions, relationships, judgment calls and bits of intuition nobody thought to document. You could easily automate the visible company while accidentally firing parts of the invisible one.
So I’m increasingly less interested in asking, “What jobs can AI replace?”
I’m more interested in: What does this company know that isn’t written down and who currently carries that knowledge?