I’m Alejandro, one of the co-founders of RSTS AI. Here’s a question we started asking about every AI employee we build, and it turned out to be the most useful one: how much of this business does it actually know?
Not how smart the model is. Every serious model is smart now. The thing that predicts whether an AI employee is useful is how much of your specific business, your customers, your process, your exceptions, it has actually learned.
Knowledge that lives in one head
Most of what a business runs on was never written down. It lives in people’s heads. One study of workplace knowledge found that 42% of institutional knowledge is unique to the individual who holds it. It isn’t in a document anywhere. It’s in the head of the person who has done the job for six years.
That’s fine until that person is out sick, or quits, or retires. Then the knowledge walks out with them, and you feel it for months. We wrote about that specific pain in when your best person leaves, the knowledge shouldn’t.
Coverage as a number
So we started treating it as something to measure. How much of the real operation has the AI employee learned? We track that as a rough coverage number, and it has turned out to be honest and useful. Low coverage, and the AI employee can only handle the obvious things. As coverage climbs, it starts handling the cases that used to need the person who “just knows.”
The number isn’t magic and it isn’t exact. But watching it go up tells us something real: the AI employee is learning your business, not just answering generic questions about your industry.
Feeding the brain
Coverage climbs the way a new hire’s does, through exposure. The difference is that what the AI employee learns doesn’t leave. It goes into a Business Brain that stays with the business: the transcripts, the documents, the way you actually handle the odd cases. This is also why a generic chat window can’t do the job. It starts every morning knowing nothing about you, which is exactly the problem we took apart in why ChatGPT forgets your business every morning.
The best question to ask about any AI employee isn’t whether it’s clever. It’s how much of your business it knows, and whether that number is going up.