anyone letting an agent phone their drivers yet?

by Hugo Duarte

Saw in yesterday's AI news roundup that Motive launched agents that act on fleet alerts themselves, calling the driver, opening a work order, that kind of thing. I plan routes for a courier firm in Porto. Half my morning is phoning drivers about stuff a system already flagged. Van running hot, street closed, wrong depot. So on paper this is my job. But the call is where you find out the van is fine and the alert is wrong, or the driver just needs ten minutes. Not sure an agent hears that. Anyone running something like this for real? Does it save time or just move the phone calls to someone else

favorite 5 comment 4

Comments

Jonas Egede 1 week ago

lol our gaffer would love this right up until it rings one of the lads at 6am about a sensor fault on a van thats been parked in the yard all week. the alarm monitoring at work went automated last year and now everyone just stops picking up. id only want it ringing when its sure but then whats the point tbf if it opened the work order and left the actual phoning to you hugo thatd be decent

Greta Nakamura 1 week ago

Hugo, Jonas basically wrote the rollout plan without meaning to. The work order and the phone call are two different permission levels, and the mistake I keep seeing is granting both on day one because the vendor demo showed both. We had the same split with paging at my last job. The agent could open and enrich tickets straight away, since a wrong ticket costs someone thirty seconds. It only got to wake a human after a few months of logged alerts where the false ones had become rare. Until then it drafted the page and a person decided. So I'd turn your question around: how often is the original fleet alert wrong? If it's one in three, an agent phoning drivers just automates the annoying calls and everyone stops picking up. Fix that rate first.

Dev Patel 6 days ago

ngl this changed my mind a bit

Hugo Duarte 6 days ago

Greta, fair question so I counted. Went back through last week's alerts on our board this morning. Roughly 40% of the overheat and fault ones were nothing, mostly two vans with a dodgy sensor. Closures were much better, maybe 1 in 10 wrong. So by your logic closures could go to an agent soon and the van alerts can't. Which is annoying because the van ones are the calls that eat my morning. Fixing the two sensors is probably the cheapest automation project we'll ever do tbh