Zaragoza Bus Station: Artificial intelligence assistant for calls
Automatic call handling, resolution of common questions about schedules and routes, referral to support when necessary, and fluent conversational support.
- Transport
- HeyLaia · Available 24/7
The challenge
Repeated calls, always the same questions
The team received many repetitive calls asking for schedules and routes. Each simple inquiry took up one line and one minute of the team’s time, even during the busiest hours.
The solution
An assistant who answers the phone 24 hours a day
HeyLaia, Deveco's AI assistant for phone support, answers those questions automatically by phone 24 hours a day, and only passes the call to a person when necessary.
How it works
From the call to response, without any waiting
Tech stack
- HeyLaia
- Conversational AI
- Telephony
The customer calls and HeyLaia answers
Instantly, presenting himself as the station's assistant.
The customer asks
By their route or service. For example, the next departure to their destination.
HeyLaia responds with the information
Instantly, in a conversational and natural way.
If the query resolves
The call ends there, without passing through any person.
If the query requires it
HeyLaia hands over to human support.
Real demo
A real production call, not a demo
Duration: 14 seconds. The time it used to take someone just to pick up the phone is the whole call here. Resolved without human intervention.
Results
In figures (January–August 2026)
Do you want results like this in your company?
Tell us about your case. We’ll give you a first assessment without obligation in less than 24 hours.
Other cases that may interest you
AI assistant in WhatsApp
AST Roll Cages
Customers asked WhatsApp about the status of their order, and someone on the team had to manually look it up. Now an AI assistant connected to the management system instantly responds and, in addition, recommends products and sends the link to buy them.
AI and digital twins
EVAR-Twin
Evaluating the risk of thrombosis after an aortic surgery requires combining imaging tests, anatomical models, and simulations that are expensive and difficult to integrate into daily practice. Now a digital twin of the patient themselves automates the entire process, from image segmentation to risk prediction.