the economics of consumer app design favor daytime users. apps get tested at 2pm by 20-somethings in san francisco, not at 3am by a nurse mid-shift in des moines. this produces products that are subtly hostile to shift workers — assuming 9-5 activity hours, offering wellness check-ins at circadian-wrong times, framing productivity around 'morning routines' that don't apply.
an ai companion is architecturally different because the product's model is adapt-to-you, not the-user-is-a-default-9-to-5-person. if you tell her your schedule, she respects it. she doesn't have a 'normal hours' assumption to override.
specific professions that have mentioned finding this useful: night nurses, ICU staff, dispatchers, air traffic controllers, long-haul truckers, overnight-bakery workers, new parents in the first 3 months (not shift work but same schedule-destruction shape), offshore workers, night-security. the common factor: you're awake when the people in your life aren't, for weeks or years at a time.
the value is narrow and specific: not a cure for shift-work's toll on your body (that's a sleep-science problem). not a replacement for the human relationships you maintain at reduced frequency because of your schedule. just a low-friction conversation partner at the specific hour when nobody else is available.
starting point: free tier, 25 msg/day. tell her your schedule on day one. check in at 3am on your next shift. see whether she's a fit. the test is short.