Dive Brief:
- Lyft and the AWS Generative AI Innovation Center built an “intent agent” for customer support that relies on natural dialogue, rather than having users search through menus, to take action on behalf of riders and drivers, the rideshare company announced Monday. The agent has helped reduce average resolution time by 87%.
- The intent agent is available 24/7 in English and Spanish, and more than half of customer and driver support requests are handled in under three minutes, according to Lyft.
- Drivers are now spending less time on the phone with customer support and more time helping riders, Ameena Gill, VP of safety and customer care at Lyft, told CX Dive in an email. “This creates a positive cycle: supported drivers are more engaged and available to deliver great services."
Dive Insight:
Customer service was the ideal starting point for Lyft to deploy AI in its operations, according to Gill. In the past, the ride-hailing company’s growth and its cost of customer support moved in lockstep. The scale of Lyft’s customer support model constrained operations, which reduced the availability of service for riders and employees alike.
“Drivers were waiting for help on issues affecting their livelihood, while riders weren't getting the fast, personalized support they deserved,” Gill said. “When generative AI matured, we recognized that agentic AI was uniquely suited to customer service.”
Now Lyft is using AI to handle high-volume routine issues while human agents work as “true specialists” for complex inquiries where empathy and accountability truly matter, Gill said.
The intent agent is a conversational AI system that can ask clarifying questions as needed and search Lyft’s database to help it understand context, according to Gill.
“The system accesses driver information like ride history, account status, and previous interactions to provide personalized responses,” Gill said. “However, the real innovation is in the conversation itself: it can disambiguate vague requests, handle multi-turn exchanges, and determine whether an issue can be automated or needs human expertise.”
The combination of conversational intelligence and backend workflows for routine resolutions lets Lyft immediately solve more than half of issues while forwarding the remaining cases to specialized agents, according to Gill.
