United StatesLogistics and transport2024 – ongoingLyft
Claude-based customer support assistant
Lyft cuts customer support time by over 87% with a Claude assistant
Problem
Riders and drivers waited 30 to 40 minutes for an agent, and agents sometimes handled three or four customers at once.
Approach
Lyft tested several models on accuracy and brand tone and chose Claude. The assistant started with driver onboarding, then moved to rider issues such as charge disputes. Safety concerns, complex disputes and sensitive cases go to human agents with an AI summary.
Results
Takeaway
The assistant started with driver onboarding and later expanded to rider issues. Safety concerns, complex disputes and sensitive cases were routed to human agents with an AI summary.
Categories
- Size
- Large enterprise
- Industry
- Logistics and transport
- Business function
- Customer service
- Innovation type
- Process innovation
- AI technology
- Generative AIConversational and voice AI
- Adoption
- Vendor solution
- Evidence
- Vendor's case study
Sources
Source summaries
Anthropic describes Lyft's Claude support assistant for riders and drivers. It reports support time down over 87% and accuracy up over 30%.
Full details are in the original sources.
