> there's no way this is cheaper than hiring an actual person.
I'm not so sure. People are pretty expensive.
Lowes are open 6am to 10pm 5 days a week and 8-8 on the weekends. That's 104 hours per week, or ~5400 hours per year.
10% downtime and you're still looking at 5000 hours per year.
Google suggests sales associates get $11.70/hour. If you add on maybe 20-30% overhead for all the associated hiring costs, taxes and extra management needed the equivalent cost would be around $75k/year. I expect the true figure would be higher.
How much are these bots? They seem to be reasonably generic, but let's say a 5 year total cost is $200k, does that sound reasonable? That's around half the cost of a real person, so if each can replace about half of a person then that's a good deal.
And that all relies on what I think are fairly low overheads, a reasonable amount of downtime for the bots, quite a high price for the bot itself and an assumption that the bots can't do anything extra that saves time (the inventory scanning for example, or collecting more useful information).
At the very least, I don't think this is obviously much more expensive, and depending on a few factors could easily be much cheaper than a real person.