AWS AI Agents For Lyft: Getting Drivers Back On The Road Faster

Seleccionar modo de aprendizaje:

Highlight:

3000 Oxford Words4000 IELTS Words5000 Oxford Words3000 Common Words1000 TOEIC Words5000 TOEFL Words

Subtítulos (112)

0:00We are [music] facing one of the

0:02biggest, I'll call it a pandemic of

0:04sorts, loneliness. And Lyft's mission is

0:07to get people off their couches and go

0:11out into the world and interact with

0:13each other.

0:18Our contact center agents, what [music]

0:20they're doing is herculean in so many

0:23ways. When a rider or a driver [music]

0:25is coming to us with something,

0:27something has gone wrong in their

0:28experience. And I think we say the best

0:30support experience is not having any

0:32support [music] experience.

0:33>> Our customer care team handles upwards

0:36of million plus contacts a month. It is

0:39high volume. It is customers who are

0:43[music] frustrated or something has

0:45happened and they need help and the

0:47agents are [music] trying their best to

0:50stay on top of that. So our goal with AI

0:53was starting really slow and building

0:55trust. They can access [music] support a

0:57few different ways, but one of the main

0:58ways will be either through the help

1:00center or through [music] their earnings

1:02tab and through there they will open up

1:04our AI assistant. And the thing that

1:06we've really emphasized there is natural

1:07language. So being able to describe

1:09[music]

1:10their issue in their own words. We've

1:12had almost 70% of our drivers today use

1:15AI. We're building [music] deep trust

1:17that if it cannot solve the issue, it

1:20very quickly will get me to human. And

1:23that's the relationship that we [music]

1:24wanted our customers and our drivers to

1:26have.

1:26>> AWS announced about two and a half years

1:29ago the AWS Genative AI innovation

1:31center and that [music] is allowing us

1:35to bring our deep science experts,

1:37engineers, strategists and [music]

1:40architects to work with our customers

1:42across the globe bringing deep technical

1:44expertise to real world problems [music]

1:46and solving them in a manner that allows

1:48businesses to get the outcomes that

1:50they're expecting. Okay, I'm a driver. I

1:52have questions about my pay. The AI

1:54assists [music] me, you know, in my own

1:56words, what am I there for?

1:58>> Some questions or or interactions will

2:00[music] be fairly straightforward. Some

2:02will be quite complex and in some

2:05[music] cases actually very ambiguous.

2:07And so that's an area where we have to

2:08work through to understand what we call

2:10intent classification. Once we get that,

2:13you then move on to doing what's called

2:15intent [music] handling. So actually

2:17executing whatever that intent was. It's

2:19really incredible to just see this

2:22magical moment where [music] they start

2:24to share their issue and they might be a

2:25little hesitant and then as we show that

2:28this is really an intelligent assistant

2:29that [music] can carry a conversation

2:31with them, you see them open up more and

2:33like share more and ask follow-up

2:35questions and then within [music]

2:37seconds I have a resolution. That

2:40experience historically would have taken

2:41[music] 10 15 minutes that AI now can

2:44solve in less than 2 minutes. So overall

2:48for our AI assistant, we're seeing usage

2:50rates around 70%. [music]

2:52So our riders and our drivers are

2:54picking this as their support channel.

2:56We are seeing really high accuracy

2:58[music] rates in terms of the

3:00resolutions that we're giving to them

3:01seeing in that like 85% 98% range.

3:04>> Technology could be amazing, but if you

3:07don't use it properly to drive either

3:09business or mission outcomes, it doesn't

3:12make it valuable. and showing the aha

3:16moment and driving those outcomes that

3:19they're after is what excites me. And as

3:21[music] we've been collaborating with

3:23the Generative AI innovation center,

3:25we've really started to transition from

3:28just being able to answer educational

3:29questions to building AI [music] agents

3:32that is conversational and natural and

3:35goes beyond just sharing a resolution

3:37and actually can take an action on

3:38behalf of the user and give them that

3:41faster resolution [music] that we are

3:42really excited about. As those low

3:45volume, low severity and [music] high

3:48volume, high complexity issues start

3:52getting [music]

3:53solved by Genai. Our humans, what we

3:57[music] are going to need them to solve,

3:59that bar of complexity is going to keep

4:01going up and up and up. And [music] so

4:04enabling them with the right knowledge

4:06and just really investing in our people,

4:09that's where [music] we're headed in the

4:10future.

Práctica de escucha

AWS AI Agents For Lyft: Getting Drivers Back On The Road Faster

Highlight: