CHAPTERS
- 0:00 – 0:30
Support capacity crisis as Lyft’s rider/driver base grew
Lyft’s support organization hit a breaking point in 2023 as demand increased faster than their existing support system could scale. Queues became overwhelmed, leading to long waits and slow issue resolution for both riders and drivers.
- •Growth in riders/drivers increased support volume
- •Existing support tooling/processes couldn’t handle demand well
- •Support queues became overwhelmed
- •Long wait times to resolve issues for riders and drivers
- 0:30 – 1:01
Evaluating AI models and choosing Claude for a more natural support experience
Lyft considered multiple AI models before selecting Claude, largely because of its “personality” and conversational tone. In real support interactions, customers responded with more organic dialogue and shared more detail about their problems.
- •Multiple AI/model options were explored
- •Claude stood out for its personality and tone
- •Customer conversations felt more organic
- •Customers opened up more about their issues, improving context
- 1:01 – 1:35
Transformational impact: dramatically faster resolutions and reinvestment in agents
After deploying Claude as an AI assistant, Lyft saw resolution time drop sharply—by 87%—turning some 30+ minute cases into seconds. The cost savings were reinvested into agent upskilling and burnout reduction, letting humans focus on high-empathy, complex issues.
- •Customer resolution time decreased by 87%
- •Some issues moved from 30+ minutes to seconds
- •Saved millions in support costs
- •Reinvested savings into agent upskilling and reducing burnout
- •Agents spend more time on cases requiring human care and empathy
