Home
Đăng nhập
Đăng ký
Loading...
Luyện nghe
Nghe
/
Video
/
Business Insider
/
AWS AI Agents For Lyft: Getting Drivers Back On The Road Faster
AWS AI Agents For Lyft: Getting Drivers Back On The Road Faster
Chọn chế độ học:
Xem phụ đề
Chọn từ
Viết lại từ
Highlight:
3000 Oxford Words
4000 IELTS Words
5000 Oxford Words
3000 Common Words
1000 TOEIC Words
5000 TOEFL Words
Phụ đề (112)
0:00
We are [music] facing one of the
0:02
biggest, I'll call it a pandemic of
0:04
sorts, loneliness. And Lyft's mission is
0:07
to get people off their couches and go
0:11
out into the world and interact with
0:13
each other.
0:18
Our contact center agents, what [music]
0:20
they're doing is herculean in so many
0:23
ways. When a rider or a driver [music]
0:25
is coming to us with something,
0:27
something has gone wrong in their
0:28
experience. And I think we say the best
0:30
support experience is not having any
0:32
support [music] experience.
0:33
>> Our customer care team handles upwards
0:36
of million plus contacts a month. It is
0:39
high volume. It is customers who are
0:43
[music] frustrated or something has
0:45
happened and they need help and the
0:47
agents are [music] trying their best to
0:50
stay on top of that. So our goal with AI
0:53
was starting really slow and building
0:55
trust. They can access [music] support a
0:57
few different ways, but one of the main
0:58
ways will be either through the help
1:00
center or through [music] their earnings
1:02
tab and through there they will open up
1:04
our AI assistant. And the thing that
1:06
we've really emphasized there is natural
1:07
language. So being able to describe
1:09
[music]
1:10
their issue in their own words. We've
1:12
had almost 70% of our drivers today use
1:15
AI. We're building [music] deep trust
1:17
that if it cannot solve the issue, it
1:20
very quickly will get me to human. And
1:23
that's the relationship that we [music]
1:24
wanted our customers and our drivers to
1:26
have.
1:26
>> AWS announced about two and a half years
1:29
ago the AWS Genative AI innovation
1:31
center and that [music] is allowing us
1:35
to bring our deep science experts,
1:37
engineers, strategists and [music]
1:40
architects to work with our customers
1:42
across the globe bringing deep technical
1:44
expertise to real world problems [music]
1:46
and solving them in a manner that allows
1:48
businesses to get the outcomes that
1:50
they're expecting. Okay, I'm a driver. I
1:52
have questions about my pay. The AI
1:54
assists [music] me, you know, in my own
1:56
words, what am I there for?
1:58
>> Some questions or or interactions will
2:00
[music] be fairly straightforward. Some
2:02
will be quite complex and in some
2:05
[music] cases actually very ambiguous.
2:07
And so that's an area where we have to
2:08
work through to understand what we call
2:10
intent classification. Once we get that,
2:13
you then move on to doing what's called
2:15
intent [music] handling. So actually
2:17
executing whatever that intent was. It's
2:19
really incredible to just see this
2:22
magical moment where [music] they start
2:24
to share their issue and they might be a
2:25
little hesitant and then as we show that
2:28
this is really an intelligent assistant
2:29
that [music] can carry a conversation
2:31
with them, you see them open up more and
2:33
like share more and ask follow-up
2:35
questions and then within [music]
2:37
seconds I have a resolution. That
2:40
experience historically would have taken
2:41
[music] 10 15 minutes that AI now can
2:44
solve in less than 2 minutes. So overall
2:48
for our AI assistant, we're seeing usage
2:50
rates around 70%. [music]
2:52
So our riders and our drivers are
2:54
picking this as their support channel.
2:56
We are seeing really high accuracy
2:58
[music] rates in terms of the
3:00
resolutions that we're giving to them
3:01
seeing in that like 85% 98% range.
3:04
>> Technology could be amazing, but if you
3:07
don't use it properly to drive either
3:09
business or mission outcomes, it doesn't
3:12
make it valuable. and showing the aha
3:16
moment and driving those outcomes that
3:19
they're after is what excites me. And as
3:21
[music] we've been collaborating with
3:23
the Generative AI innovation center,
3:25
we've really started to transition from
3:28
just being able to answer educational
3:29
questions to building AI [music] agents
3:32
that is conversational and natural and
3:35
goes beyond just sharing a resolution
3:37
and actually can take an action on
3:38
behalf of the user and give them that
3:41
faster resolution [music] that we are
3:42
really excited about. As those low
3:45
volume, low severity and [music] high
3:48
volume, high complexity issues start
3:52
getting [music]
3:53
solved by Genai. Our humans, what we
3:57
[music] are going to need them to solve,
3:59
that bar of complexity is going to keep
4:01
going up and up and up. And [music] so
4:04
enabling them with the right knowledge
4:06
and just really investing in our people,
4:09
that's where [music] we're headed in the
4:10
future.
