 | TL;DR: Nick Saraev ran the numbers on AI voice agents — a 1% chance a customer clocks it as fake can cost 20-40% of your total revenue, not the 1% you thought you were risking.
That gap between what people think they're risking and what they're actually risking is the whole game. Every trades or home-services owner getting pitched "full automation" right now is being sold the 1% number, never the 40% one.
The fix Nick lays out isn't rejecting AI — it's AI that identifies itself, buys your team 15-20 seconds, then hands off. Never pretending to be something it isn't.
We live in the upper floors of a high-rise apartment. I remember many years ago, my 5yo son and 3yo daughter would bolt out of the barely opened elevator door, and race each other down the bridge-corridor towards our unit. My son was just about the same height as the corridor's parapet walls and grill-railing. And my daughter was just game for anything following him. Of course I was horrified. What if they tripped over the railing... Unthinkable. So I yelled at them, "不要跑！慢慢走！" (Don't run! Walk slowly!). And they did. After a few more rounds of my yelling and them complying over time – they got the picture. The more I use AI, the more I feel like they're just toddlers – needed yelling (proper framework-prompting). —————————————— You already know the math your AI vendor pitch won't show you: what one bad call costs against what automation actually saves.
Curious where other owner-operators land on this — has anyone actually run the "buys you 15-20 seconds" hybrid in practice, or is it still mostly vendor pitch? Drop your take below.
Clip credit: Nick Saraev — full video on his channel, Nick Saraev Unfiltered. DM for credit or removal requests. [link] [comments] |
Predictive technology: computational social science
