Case study · Paratransit call center
A paratransit call center booked 1,425 rides with AI in its first 12 weeks
A paratransit call center runs AiPRL on its ride-booking and information lines in English and Spanish. From August 3 to October 27, 2026, AiPRL handled 9,146 calls, resolved 35% with no person involved and booked 1,425 rides, 721 of them by the AI alone.
- 1,425
- rides booked, 721 by the AI alone and 704 started by the AI before a handoff
- 35%
- of 9,146 calls resolved with zero human touch
- 2m 00s
- average AI call, against 5m 37s after a handoff
- 95%
- of 8,520 scored calls ended neutral or better
Window: Aug 3 – Oct 27, 2026. Source: AiPRL executive performance report, Oct 27, 2026.
The problem
Riders call to book, confirm or change trips and wait on hold while staff juggle the phones. Calls surge when the booking window opens, and a missed confirmation becomes a missed ride.
What we did
Put AiPRL in front of the ride-booking and information lines, in English and Spanish.
Connected trip booking, so rides are booked and confirmed on the call.
Every call the AI doesn't finish goes to an agent with the conversation attached; reminders and confirmations go out by text.
Where every call ends
Source: AiPRL executive performance report, Aug 3 – Oct 27, 2026; percentage of 9,146 calls.
Of 5,123 transfers, 2,817 were the rider choosing a person and 2,226 followed the center's routing rules; 79 were the AI unable to finish, under 1% of calls. Spanish calls were resolved with no agent 44% of the time, against 31% in English. 931 reminder and confirmation texts went out.