WEBVTT - ICYMI: Facial Recognition vs Facial Verification

0:00:02.520 --> 0:00:09.560
<v Speaker 1>Bloomberg Audio Studios, Podcasts, radio news. You're listening to Bloomberg

0:00:09.640 --> 0:00:15.040
<v Speaker 1>Business Week with Carol Masser and tim Stenoveek on Bloomberg Radio. Carol,

0:00:15.120 --> 0:00:18.520
<v Speaker 1>you have traveled abroad in the last year, did it? Yeah?

0:00:18.560 --> 0:00:19.840
<v Speaker 1>You did. You went and visited your daughter.

0:00:20.040 --> 0:00:20.920
<v Speaker 2>Oh yeah, that's right.

0:00:21.000 --> 0:00:22.520
<v Speaker 1>Yeah, he came back to the airport.

0:00:22.600 --> 0:00:22.960
<v Speaker 2>Thanks too.

0:00:23.079 --> 0:00:25.520
<v Speaker 1>Yeah, no, prom he came back to the airport.

0:00:25.720 --> 0:00:26.000
<v Speaker 2>Yeah.

0:00:26.040 --> 0:00:29.240
<v Speaker 1>You're in line Customs and Border Protection totally. You go

0:00:29.360 --> 0:00:31.680
<v Speaker 1>up to them, you give them your passport. They said,

0:00:31.680 --> 0:00:34.080
<v Speaker 1>you say, I'm Carol Masser, and they say, how do

0:00:34.120 --> 0:00:38.680
<v Speaker 1>we know you're Carol Masser? Facial recognition, well, facial verification.

0:00:39.920 --> 0:00:42.360
<v Speaker 1>There's a difference. Andrew Budd is going to explain it.

0:00:42.479 --> 0:00:45.479
<v Speaker 1>And Andrew Budd's product is actually used by Customs and

0:00:45.560 --> 0:00:49.320
<v Speaker 1>Border Protection in that exact situation to verify that we

0:00:49.520 --> 0:00:52.040
<v Speaker 1>are who we say we are when we are crossing borders.

0:00:52.120 --> 0:00:54.520
<v Speaker 2>I love that you said that because it's an interesting designation.

0:00:54.680 --> 0:00:58.040
<v Speaker 2>So let's get to it. Alex, I'm sorry. Andrew Bud

0:00:58.200 --> 0:01:01.000
<v Speaker 2>is with us. He's founder and CEO of I Prove

0:01:01.440 --> 0:01:04.440
<v Speaker 2>and he joins us from Amsterdam. It's good to have

0:01:04.520 --> 0:01:08.360
<v Speaker 2>you here, especially as we are continuing to embark on

0:01:08.400 --> 0:01:11.160
<v Speaker 2>our new world order thanks to AI. Tell us a

0:01:11.160 --> 0:01:13.480
<v Speaker 2>little bit more about your company and the importance of

0:01:13.520 --> 0:01:18.039
<v Speaker 2>this distinction between facial verification versus facial recognition, because they're

0:01:18.160 --> 0:01:18.959
<v Speaker 2>very different things.

0:01:19.480 --> 0:01:21.399
<v Speaker 3>They are very different, and it's very easy for them

0:01:21.440 --> 0:01:25.479
<v Speaker 3>to be confused. Look, facial verification is about empowering the person,

0:01:25.560 --> 0:01:29.600
<v Speaker 3>empowering the citizen. If you can use your face to

0:01:29.760 --> 0:01:32.560
<v Speaker 3>secure your own identity, if you know that it's happening,

0:01:32.840 --> 0:01:38.160
<v Speaker 3>if you've consented to it's happening, if you get personal

0:01:38.240 --> 0:01:42.119
<v Speaker 3>benefit from it's happening, and your privacy is protected, that's

0:01:42.200 --> 0:01:46.160
<v Speaker 3>facial verification. That's empowering the citizen. Facial recognition is about

0:01:46.240 --> 0:01:51.040
<v Speaker 3>identifying people, often without their consent, often without their knowledge.

0:01:51.080 --> 0:01:55.360
<v Speaker 3>It's a completely different thing, and that's about empowering organizations.

0:01:55.360 --> 0:01:59.920
<v Speaker 3>It's empowering, about tracking people. Facial verification is about empowering

0:02:00.600 --> 0:02:03.600
<v Speaker 3>It's about enabling you to use your face as a

0:02:04.120 --> 0:02:07.840
<v Speaker 3>security credential. And it's a great chrisk security credential because

0:02:07.840 --> 0:02:09.400
<v Speaker 3>it can't be stolen, it can't be lost, and it

0:02:09.400 --> 0:02:10.000
<v Speaker 3>can't be shared.

0:02:10.160 --> 0:02:12.880
<v Speaker 1>So is a face ID on an iPhone, for example?

0:02:12.919 --> 0:02:15.200
<v Speaker 1>Is that facial recognition or facial verification.

0:02:16.400 --> 0:02:19.239
<v Speaker 3>That's facial verification very much. Facial verification.

0:02:19.400 --> 0:02:24.560
<v Speaker 1>Facial recognition is essentially used for security purposes by third parties.

0:02:24.600 --> 0:02:26.680
<v Speaker 1>Then how would you define facial recognition?

0:02:27.600 --> 0:02:33.000
<v Speaker 3>Yes, offacial recognition is really, in our terms about surveillance.

0:02:33.120 --> 0:02:36.959
<v Speaker 3>It's about identifying who the person is. Facial verification is

0:02:36.960 --> 0:02:38.800
<v Speaker 3>when I start by saying, Hey, I'm the owner of

0:02:38.840 --> 0:02:41.320
<v Speaker 3>this phone. Hey, I'm the owner of this passport. Hey

0:02:41.360 --> 0:02:44.240
<v Speaker 3>I'm the owner of this of this enterprise account. And

0:02:44.320 --> 0:02:46.320
<v Speaker 3>I want to prove I now want to prove my face.

0:02:46.520 --> 0:02:48.960
<v Speaker 3>I want to use my face to prove that I

0:02:49.000 --> 0:02:52.079
<v Speaker 3>am who I've claimed to be. But with facial verification,

0:02:52.240 --> 0:02:55.279
<v Speaker 3>I start by claiming who I am. In facial recognition,

0:02:55.360 --> 0:02:56.680
<v Speaker 3>the system recognized.

0:02:57.360 --> 0:03:01.639
<v Speaker 2>When my doorbell camera says he I recognize somebody familiar

0:03:01.639 --> 0:03:03.720
<v Speaker 2>and it's the UPS delivery person. And then it says,

0:03:04.000 --> 0:03:05.880
<v Speaker 2>I don't know it's your husband or it's your wife.

0:03:06.760 --> 0:03:10.640
<v Speaker 2>Is that facial recognition gone wrong? In other words, like

0:03:10.680 --> 0:03:13.880
<v Speaker 2>it's picked up some cues and there's some familiar you know,

0:03:14.200 --> 0:03:15.240
<v Speaker 2>like I try to understand this.

0:03:15.919 --> 0:03:18.480
<v Speaker 3>Yeah, So ask yourself always does the does does the

0:03:18.520 --> 0:03:22.360
<v Speaker 3>subject have agency? Does the subject, Does the subject know

0:03:22.400 --> 0:03:24.480
<v Speaker 3>it's happening, have they consented to it happening? And are

0:03:24.480 --> 0:03:28.200
<v Speaker 3>they getting personal benefit from it happening? Those are the

0:03:28.240 --> 0:03:30.000
<v Speaker 3>real questions. So when you turn up at the border

0:03:30.000 --> 0:03:35.120
<v Speaker 3>with CBP, we have systems installed in nearly a dozen

0:03:35.640 --> 0:03:39.000
<v Speaker 3>airports around the US for global entry and now increases

0:03:39.000 --> 0:03:42.120
<v Speaker 3>in the also for US citizens at Orlando, for example,

0:03:42.600 --> 0:03:47.040
<v Speaker 3>you turn up, you can choose to go through these

0:03:47.080 --> 0:03:51.840
<v Speaker 3>these these systems that are based upon facial verification. You

0:03:51.880 --> 0:03:54.920
<v Speaker 3>are searching your agency by choosing to go uh and

0:03:55.280 --> 0:03:57.160
<v Speaker 3>go to to go and present yourself. You don't have to,

0:03:57.880 --> 0:03:59.920
<v Speaker 3>but if you do, it's a lot faster, it's a

0:04:00.120 --> 0:04:02.680
<v Speaker 3>lot simpler, it's a much better experience, it's a lot

0:04:02.720 --> 0:04:03.320
<v Speaker 3>more convenient.

0:04:03.600 --> 0:04:05.240
<v Speaker 1>Well, let's talk about a little bit up the big

0:04:05.280 --> 0:04:08.720
<v Speaker 1>business in the solution, because as I mentioned, the technology

0:04:08.720 --> 0:04:10.920
<v Speaker 1>of I prove is being used by customs and Border

0:04:10.920 --> 0:04:15.560
<v Speaker 1>Protection at entryways at airports. For example, You've got about

0:04:15.600 --> 0:04:18.080
<v Speaker 1>you've raised about seventy million dollars. You've got employees all

0:04:18.120 --> 0:04:20.880
<v Speaker 1>over the world, the US, UK, Latin America, Asia, Pacific,

0:04:21.200 --> 0:04:23.520
<v Speaker 1>and more. How does the technology work?

0:04:24.560 --> 0:04:28.719
<v Speaker 3>So most of what we do actually involves Verifying that

0:04:28.720 --> 0:04:30.680
<v Speaker 3>people are whom they claim to be when they're not

0:04:30.720 --> 0:04:33.479
<v Speaker 3>physically doesn't when they're at home, on their couch, for example,

0:04:33.680 --> 0:04:36.880
<v Speaker 3>and they're trying to assert their identity for the purposes

0:04:36.960 --> 0:04:41.480
<v Speaker 3>of opening an account or maybe very crucially resetting their credentials.

0:04:41.480 --> 0:04:44.320
<v Speaker 3>When something's gone wrong. When a person's on their couch,

0:04:44.600 --> 0:04:46.760
<v Speaker 3>it's really very difficult to know whether they are whom

0:04:46.760 --> 0:04:48.440
<v Speaker 3>they claim to be. You can look at them. You

0:04:48.440 --> 0:04:51.480
<v Speaker 3>can use your mobile phone to stream video to a

0:04:51.560 --> 0:04:54.279
<v Speaker 3>central system and check do they look who like they should?

0:04:54.760 --> 0:04:56.279
<v Speaker 3>That they look like who they claim to be, But

0:04:56.400 --> 0:04:58.520
<v Speaker 3>how do you know they real? They're real? We live

0:04:58.560 --> 0:05:01.880
<v Speaker 3>in a world of AI of deep When it's extremely

0:05:01.920 --> 0:05:07.240
<v Speaker 3>straightforward to put a digital mosque over over a person's

0:05:07.240 --> 0:05:12.040
<v Speaker 3>face and pretend to be to impersonate somebody, we stop

0:05:12.120 --> 0:05:16.560
<v Speaker 3>that impersonation. We can detect when very accurately, when a

0:05:16.560 --> 0:05:18.560
<v Speaker 3>deep fake is trying to be used to impersonate someone.

0:05:18.640 --> 0:05:20.960
<v Speaker 2>Come on, Andrew, we've all seen like the Mission Impossible

0:05:21.000 --> 0:05:22.000
<v Speaker 2>movies or something like.

0:05:22.160 --> 0:05:24.640
<v Speaker 1>Those deep faces that's a long time ago too, are

0:05:24.720 --> 0:05:27.120
<v Speaker 1>used twenty years ago, thirty years ago. I use nowadays.

0:05:27.160 --> 0:05:30.200
<v Speaker 2>It's a lot better, right, exactly like the technology just

0:05:30.200 --> 0:05:33.280
<v Speaker 2>gets better and better. So I mean, how do we

0:05:33.320 --> 0:05:38.279
<v Speaker 2>make sure the facial verification systems stay ahead of you know,

0:05:38.560 --> 0:05:41.520
<v Speaker 2>the AI that is creating those deep fakes in terms

0:05:41.520 --> 0:05:45.080
<v Speaker 2>of their complexity. How do you do that?

0:05:45.080 --> 0:05:47.839
<v Speaker 3>That's the central problem that we solve. That's what we've

0:05:47.839 --> 0:05:49.880
<v Speaker 3>built a global business doing, and we've been doing it

0:05:49.920 --> 0:05:52.560
<v Speaker 3>now for quite some time. Two things we do. One

0:05:52.600 --> 0:05:55.600
<v Speaker 3>is we have we change the physics of the situation.

0:05:55.720 --> 0:05:58.719
<v Speaker 3>We illuminate the user's face with a rapidly changing sequence

0:05:58.720 --> 0:06:01.600
<v Speaker 3>of colors from screen of their own device. It's an

0:06:01.680 --> 0:06:06.000
<v Speaker 3>unpredictable sequence of colors and it illuminates their face. And

0:06:06.080 --> 0:06:08.640
<v Speaker 3>while those colors are illuminating their face, we stream video

0:06:08.680 --> 0:06:11.200
<v Speaker 3>back to our servers where we look at how those

0:06:11.320 --> 0:06:15.000
<v Speaker 3>that screen illumination. There's changing colors reflect off the user's face.

0:06:15.040 --> 0:06:17.120
<v Speaker 3>They're unpredictable. The attackers can't know what it's going to

0:06:17.160 --> 0:06:20.080
<v Speaker 3>look like, so they can't pre prepare some beautiful deep fake.

0:06:20.240 --> 0:06:22.159
<v Speaker 3>That's one element of it. And then the other is

0:06:22.320 --> 0:06:26.520
<v Speaker 3>we analyze everything that's happening worldwide all the time. We

0:06:26.600 --> 0:06:28.800
<v Speaker 3>filter the systems and we study it so we can

0:06:28.839 --> 0:06:31.799
<v Speaker 3>detect the attacks. We can detect if a foreign power,

0:06:31.800 --> 0:06:34.320
<v Speaker 3>for example, is busily mounting a set of experiments, and

0:06:34.360 --> 0:06:36.960
<v Speaker 3>we learn from them, and we learn more from them

0:06:37.120 --> 0:06:39.760
<v Speaker 3>than they learn from us. So we stay ahead because

0:06:39.760 --> 0:06:43.120
<v Speaker 3>we have better. We have visibility and they have no visibility.

0:06:43.200 --> 0:06:45.640
<v Speaker 3>You know, information advantage enables us to win.

0:06:45.839 --> 0:06:47.800
<v Speaker 1>Carol and I were talking ahead of this when we

0:06:47.800 --> 0:06:49.600
<v Speaker 1>were preparing for this, and we've been traveling a lot,

0:06:49.640 --> 0:06:52.440
<v Speaker 1>so it's on our mind. But we were members of

0:06:52.480 --> 0:06:54.960
<v Speaker 1>Clear now because we are on the plane a lot.

0:06:55.720 --> 0:06:59.719
<v Speaker 1>Sometimes it's faster than the traditional security line. Sometimes it's not.

0:07:00.000 --> 0:07:02.280
<v Speaker 1>It depends on that, Yeah, depending on where you are.

0:07:02.440 --> 0:07:05.240
<v Speaker 1>But just sometimes we can get through security without even

0:07:05.839 --> 0:07:10.240
<v Speaker 1>showing our plastic IDs. Does Clear use a system such

0:07:10.240 --> 0:07:10.920
<v Speaker 1>as I prove?

0:07:12.040 --> 0:07:15.120
<v Speaker 3>I can't comment on on clear. Clear have their own technology,

0:07:15.120 --> 0:07:17.760
<v Speaker 3>they have a very sophisticated technology. What I can say

0:07:18.040 --> 0:07:20.679
<v Speaker 3>is that when you go through Global Entry in places

0:07:20.720 --> 0:07:23.920
<v Speaker 3>like Newark or Los Angeles or Miami, you walk up

0:07:23.960 --> 0:07:26.920
<v Speaker 3>to it and I prove terminal, you don't do anything.

0:07:26.920 --> 0:07:28.760
<v Speaker 3>You don't have to put out any card or any

0:07:28.800 --> 0:07:33.400
<v Speaker 3>identification system. Your face is. Your face is captured and

0:07:33.440 --> 0:07:37.280
<v Speaker 3>it's matched by a CBP itself against their traveler verification

0:07:37.440 --> 0:07:40.440
<v Speaker 3>system they have, You've already in the past given them

0:07:40.520 --> 0:07:44.440
<v Speaker 3>all of your details so that you're expected because you

0:07:44.520 --> 0:07:47.120
<v Speaker 3>are coming in on a particular flight, and they kind

0:07:47.160 --> 0:07:50.880
<v Speaker 3>of go, oh, yes, that's Andrew budd We were expecting him.

0:07:51.440 --> 0:07:53.560
<v Speaker 3>We know we know his details because he's given them

0:07:53.560 --> 0:07:56.080
<v Speaker 3>to us and we had them already. He looks like

0:07:56.120 --> 0:07:58.920
<v Speaker 3>he's supposed to look like. Go forward. And the rate

0:07:58.920 --> 0:08:01.200
<v Speaker 3>at which people can be pro like that is awesome.

0:08:01.640 --> 0:08:04.440
<v Speaker 3>The places where we've been installed, we've eliminated. We've eliminated

0:08:04.480 --> 0:08:05.800
<v Speaker 3>most of the cues or what happens.

0:08:06.160 --> 0:08:08.440
<v Speaker 1>What happens if I don't look like I looked in

0:08:08.440 --> 0:08:11.280
<v Speaker 1>the photo or the documents that I provided to the

0:08:11.320 --> 0:08:14.080
<v Speaker 1>government maybe five or ten years ago. What if I've aged?

0:08:14.120 --> 0:08:16.920
<v Speaker 1>What if I've have a different hair collar, What if

0:08:16.960 --> 0:08:18.840
<v Speaker 1>I decided to get up saying what if I had

0:08:18.920 --> 0:08:20.320
<v Speaker 1>a lot of work done to my face.

0:08:21.240 --> 0:08:24.280
<v Speaker 3>One of the great advantages of face verification technology is

0:08:24.280 --> 0:08:27.000
<v Speaker 3>that it works extraordinarily well, and it works a lot

0:08:27.040 --> 0:08:30.200
<v Speaker 3>better than people do. There was a study done about

0:08:30.720 --> 0:08:33.920
<v Speaker 3>a number of about a decade ago by university which

0:08:33.960 --> 0:08:38.840
<v Speaker 3>showed that the performance of really skilled passport officers was

0:08:38.840 --> 0:08:41.520
<v Speaker 3>of a level that is now about one hundred thousand

0:08:41.520 --> 0:08:45.760
<v Speaker 3>times less good than a modern face verification system. Face

0:08:45.880 --> 0:08:50.079
<v Speaker 3>verification systems tend to be relatively indifferent to face furniture

0:08:50.120 --> 0:08:53.280
<v Speaker 3>as we call it, beards, glasses, piercings, and so on.

0:08:54.280 --> 0:08:57.760
<v Speaker 3>They're really very good at matching people.

0:08:59.120 --> 0:09:00.880
<v Speaker 2>Hey, one day, when I just ask you, before we

0:09:00.880 --> 0:09:02.560
<v Speaker 2>get a little bit more into the business, just quickly,

0:09:02.760 --> 0:09:05.600
<v Speaker 2>is a face better than a fingerprint?

0:09:07.040 --> 0:09:12.120
<v Speaker 3>Yes? Absolutely? Why because they solve two different problems. When

0:09:12.160 --> 0:09:14.560
<v Speaker 3>you're trying to verify somebody, you want to make sure

0:09:14.559 --> 0:09:17.400
<v Speaker 3>that there's a good match, and you're not trying to

0:09:17.600 --> 0:09:19.880
<v Speaker 3>figure out whether that which of seven billion people in

0:09:19.920 --> 0:09:23.760
<v Speaker 3>the world this is. You know, they're expecting you and

0:09:23.960 --> 0:09:26.000
<v Speaker 3>the challenges, and you just have to make sure that

0:09:26.080 --> 0:09:28.840
<v Speaker 3>you look like the person accurately. There is so and

0:09:28.880 --> 0:09:30.559
<v Speaker 3>a face is extremely good at doing that. But the

0:09:30.559 --> 0:09:32.800
<v Speaker 3>reason a face is better is the real way that

0:09:32.840 --> 0:09:37.520
<v Speaker 3>these things get attacked, especially remotely, is by fate forgeries, copies.

0:09:38.000 --> 0:09:41.160
<v Speaker 3>These things. These things are and it's much easier to

0:09:41.200 --> 0:09:43.640
<v Speaker 3>detect a It's much easier and much more feasible for

0:09:43.679 --> 0:09:48.200
<v Speaker 3>us to detect a forged or deep faked face or

0:09:48.240 --> 0:09:51.160
<v Speaker 3>deep faked fingerprint because there's so much more information, there's

0:09:51.160 --> 0:09:53.800
<v Speaker 3>so much more texture, there's so much more depth, there's

0:09:53.800 --> 0:09:58.040
<v Speaker 3>so much more information around a face. The ambient illumination

0:09:58.200 --> 0:10:00.559
<v Speaker 3>is very important, so you get much more more information

0:10:00.600 --> 0:10:03.160
<v Speaker 3>from a face, which makes it much more reliable to

0:10:03.320 --> 0:10:05.080
<v Speaker 3>protect whether a face is real or not.

0:10:05.320 --> 0:10:07.200
<v Speaker 1>Interesting, are we going to start to see this type

0:10:07.240 --> 0:10:10.400
<v Speaker 1>of technology being used in more places than just unlocking

0:10:10.440 --> 0:10:12.960
<v Speaker 1>something on our phone or getting through the line at

0:10:12.960 --> 0:10:14.960
<v Speaker 1>the airport where.

0:10:14.880 --> 0:10:20.160
<v Speaker 3>Absolutely you're going to see this so so already. If

0:10:20.200 --> 0:10:22.120
<v Speaker 3>you want to set up bank accounts in many parts

0:10:22.160 --> 0:10:25.280
<v Speaker 3>of the world, you go through you I prove yourself remotely,

0:10:25.400 --> 0:10:28.000
<v Speaker 3>You I prove yourself to do the Know your Customer

0:10:28.080 --> 0:10:30.240
<v Speaker 3>KYC thing to make sure that the person setting up

0:10:30.320 --> 0:10:33.040
<v Speaker 3>the account is genuinely the person that they claim to

0:10:33.040 --> 0:10:36.520
<v Speaker 3>be and not somebody being a money an impersonating money

0:10:36.600 --> 0:10:39.840
<v Speaker 3>mule who's going to launder money. That's already happening in

0:10:39.880 --> 0:10:41.600
<v Speaker 3>many parts of the world, not so much in the

0:10:41.679 --> 0:10:45.959
<v Speaker 3>US yet. For enterprises, we're going to see staff doing

0:10:46.000 --> 0:10:48.920
<v Speaker 3>this both at the time when they're hired and also

0:10:49.160 --> 0:10:51.680
<v Speaker 3>when they have to reset their credentials when they're hired.

0:10:51.880 --> 0:10:54.440
<v Speaker 3>There's there's now a real problem in the United States

0:10:54.520 --> 0:10:59.520
<v Speaker 3>with impersonators, particularly members of the North Korean Secret Service,

0:11:00.040 --> 0:11:04.400
<v Speaker 3>impersonating American staff and being signed up and given jobs

0:11:04.440 --> 0:11:07.040
<v Speaker 3>to work remotely. The first of the there was a

0:11:07.080 --> 0:11:09.000
<v Speaker 3>very public case in the summer of last year. Now,

0:11:09.240 --> 0:11:11.640
<v Speaker 3>the CITO of Mandian, which is a cybersecurity company, says

0:11:11.640 --> 0:11:15.640
<v Speaker 3>that literally every fortune five hundred company has at least dozens,

0:11:15.679 --> 0:11:19.280
<v Speaker 3>if not hundreds of job applications North Korean IT workers

0:11:19.360 --> 0:11:23.280
<v Speaker 3>whose faces have been deep faked to look like American

0:11:23.320 --> 0:11:26.960
<v Speaker 3>staff when they're not. Three hundred US companies have been

0:11:27.040 --> 0:11:31.400
<v Speaker 3>scammed by a woman who played guilty in February for

0:11:31.480 --> 0:11:34.520
<v Speaker 3>having run a scheme like this. So it's a huge thing.

0:11:34.600 --> 0:11:37.360
<v Speaker 3>So our technology will make sure that the person who

0:11:37.440 --> 0:11:40.680
<v Speaker 3>is being hired remotely is whom they claim to be.

0:11:40.720 --> 0:11:43.000
<v Speaker 3>And then you know when you come to reset your password,

0:11:43.360 --> 0:11:46.040
<v Speaker 3>you haven't got any other way to assure yourself all right,

0:11:46.440 --> 0:11:47.080
<v Speaker 3>that will be used for that.

0:11:47.840 --> 0:11:50.240
<v Speaker 2>Stay in touch, Love to check in with you again

0:11:50.280 --> 0:11:53.160
<v Speaker 2>in the future. Andrew bad founder and CEO of I

0:11:53.360 --> 0:11:55.360
<v Speaker 2>Prove joining us right here on Business Week