WEBVTT - Adobe Launches Adobe Foundry Services in Next Big AI Move

0:00:02.520 --> 0:00:09.520
<v Speaker 1>Bloomberg Audio Studios, Podcasts, radio news. This is Bloomberg Business

0:00:09.560 --> 0:00:13.840
<v Speaker 1>Week with Carol Masser and Tim Steneveek on Bloomberg Radio.

0:00:14.480 --> 0:00:17.600
<v Speaker 2>Carol, have you played around with open AIS? Sora?

0:00:17.840 --> 0:00:19.720
<v Speaker 3>Well, I'm gonna be quite honest with you.

0:00:19.880 --> 0:00:22.440
<v Speaker 2>I wrote this, Okay, to be honest, that.

0:00:22.400 --> 0:00:24.720
<v Speaker 3>Would be no, because I'm an honest kind of g hell.

0:00:25.640 --> 0:00:26.920
<v Speaker 3>But you just kind of shared it with me.

0:00:27.000 --> 0:00:29.200
<v Speaker 2>I just invited you, I inte because I've heard about

0:00:29.200 --> 0:00:31.040
<v Speaker 2>it with you. Got five more people if you want,

0:00:31.680 --> 0:00:34.120
<v Speaker 2>if you're interested in creating some AI slop, what do

0:00:34.159 --> 0:00:36.080
<v Speaker 2>you have to be invited? I think this is just

0:00:36.080 --> 0:00:37.839
<v Speaker 2>what they do to control the number of people who

0:00:37.840 --> 0:00:39.400
<v Speaker 2>sign up for it in the beginning, because it's so

0:00:40.000 --> 0:00:41.839
<v Speaker 2>energy intensive. But I don't know exactly what right.

0:00:41.880 --> 0:00:44.720
<v Speaker 3>Well, tell everybody, Tell tell them Tim, what's behind the

0:00:44.760 --> 0:00:45.479
<v Speaker 3>big doors.

0:00:45.600 --> 0:00:47.519
<v Speaker 2>So I bring this up because Adobe is doing its

0:00:47.520 --> 0:00:50.920
<v Speaker 2>own thing with AI generated video and like Firefly for images.

0:00:50.960 --> 0:00:52.879
<v Speaker 2>What you remember from Adobe, We've talked about that for

0:00:52.880 --> 0:00:55.080
<v Speaker 2>a couple of years now. This is supposed to be

0:00:55.120 --> 0:00:57.720
<v Speaker 2>brand and copyright safe. The company out with this new

0:00:57.720 --> 0:01:01.200
<v Speaker 2>product is called Adobe AI Foundry, and it's essentially a

0:01:01.240 --> 0:01:04.040
<v Speaker 2>model that lets companies build their own GENAI models that

0:01:04.080 --> 0:01:08.160
<v Speaker 2>are trained on their own material. Hannah el Soccer is

0:01:08.280 --> 0:01:10.760
<v Speaker 2>vice president of Jenai New Business Ventures at Adobe. She

0:01:10.800 --> 0:01:13.200
<v Speaker 2>joins us from Queen's Hannah, Did I get that right?

0:01:14.840 --> 0:01:17.800
<v Speaker 4>You got that right? Okay in Queens and I'm with

0:01:17.840 --> 0:01:18.640
<v Speaker 4>Adobe as well.

0:01:19.360 --> 0:01:22.440
<v Speaker 2>So is that? But that makes sense? Right? Like the

0:01:22.760 --> 0:01:25.080
<v Speaker 2>way that you guys think about this is you create

0:01:25.760 --> 0:01:30.600
<v Speaker 2>you can use prompts to create copyright safe videos that

0:01:31.120 --> 0:01:35.679
<v Speaker 2>companies enterprises can use in marketing and in press materials. Correct.

0:01:36.720 --> 0:01:40.200
<v Speaker 4>Yeah, so AI. You know, AI has been quite the

0:01:40.200 --> 0:01:43.120
<v Speaker 4>whirlwind these last couple of years, and Adobe AI foundry

0:01:43.560 --> 0:01:48.480
<v Speaker 4>really allows fortune two thousand enterprises to deeply tune, responsibly

0:01:48.520 --> 0:01:51.360
<v Speaker 4>trained A models rooted in their own brand and IP.

0:01:51.760 --> 0:01:55.160
<v Speaker 4>So we care a lot about IP protection on brand creation,

0:01:55.240 --> 0:01:58.920
<v Speaker 4>as you mentioned across image video, three D vector audio,

0:01:59.000 --> 0:01:59.760
<v Speaker 4>all those things.

0:02:00.320 --> 0:02:04.000
<v Speaker 2>So what's the use case that you envision for these companies.

0:02:04.320 --> 0:02:06.120
<v Speaker 2>I think a lot of people are having trouble understanding

0:02:06.120 --> 0:02:07.480
<v Speaker 2>this new world, So you've got to break it down

0:02:07.480 --> 0:02:09.640
<v Speaker 2>for us in a way that's like, Okay, you're telling

0:02:09.680 --> 0:02:11.680
<v Speaker 2>your friend about it, who doesn't even know what you

0:02:11.720 --> 0:02:13.520
<v Speaker 2>do at Adobe, of.

0:02:13.480 --> 0:02:15.680
<v Speaker 4>Course, and you know, a part of it is that

0:02:15.880 --> 0:02:19.120
<v Speaker 4>we are in an attention economy, and so the amount

0:02:19.200 --> 0:02:24.800
<v Speaker 4>of media that we all consume now, whether it's audio, radio, video, shorts,

0:02:25.280 --> 0:02:28.960
<v Speaker 4>is very high. And the key is being able to

0:02:29.440 --> 0:02:33.640
<v Speaker 4>represent the brand, represent the franchise in a way that's

0:02:33.680 --> 0:02:37.080
<v Speaker 4>responsible and on brand. That just takes a lot of

0:02:37.120 --> 0:02:39.840
<v Speaker 4>effort and a lot of time. And so the idea

0:02:40.000 --> 0:02:43.680
<v Speaker 4>of Adobe AI Foundry is that once we responsibly train

0:02:43.800 --> 0:02:47.639
<v Speaker 4>these brands where you protect your own IP, they can

0:02:47.680 --> 0:02:51.919
<v Speaker 4>be used in the Adobe products that marketers creatives are using,

0:02:52.040 --> 0:02:56.560
<v Speaker 4>like gen Studio, like Photoshop, firefly boards, Premiere Pro. These

0:02:56.600 --> 0:02:59.600
<v Speaker 4>are all part of the workflow in a creative and

0:02:59.639 --> 0:03:02.880
<v Speaker 4>marketing company. So it gives you trust, gives you precision

0:03:02.880 --> 0:03:03.520
<v Speaker 4>and creativity.

0:03:03.639 --> 0:03:07.000
<v Speaker 2>So let's say a company uses the product that you're offering.

0:03:07.040 --> 0:03:09.120
<v Speaker 2>It's brand safe, they trained it on their own data,

0:03:09.160 --> 0:03:11.880
<v Speaker 2>they trained it on their own materials, they put it

0:03:11.880 --> 0:03:15.200
<v Speaker 2>out in the world. Then that becomes part of the

0:03:15.240 --> 0:03:20.600
<v Speaker 2>ecosystem that maybe some other model is trained on and

0:03:20.919 --> 0:03:23.280
<v Speaker 2>it's kind of on the loose. Now, is that possible.

0:03:24.520 --> 0:03:26.480
<v Speaker 4>I don't think that's the use. That's a great question.

0:03:26.520 --> 0:03:29.480
<v Speaker 4>I don't think that's exactly the use case where no, I'm.

0:03:29.320 --> 0:03:30.720
<v Speaker 2>Not saying it's the use case. I'm saying it's a

0:03:30.720 --> 0:03:35.360
<v Speaker 2>consequence of content being out in the wild whether we

0:03:35.440 --> 0:03:40.520
<v Speaker 2>like it or not. What we do is training AI models. Yeah.

0:03:40.560 --> 0:03:42.880
<v Speaker 4>The way we think about it is these are enterprise

0:03:42.960 --> 0:03:46.120
<v Speaker 4>grade and IP protected and so therefore the use of

0:03:46.360 --> 0:03:50.680
<v Speaker 4>the leaders like the Home Depot Disney, they're co developing

0:03:50.720 --> 0:03:53.360
<v Speaker 4>this with us for their own use. This is not

0:03:53.480 --> 0:03:56.120
<v Speaker 4>about putting these models out in the wild per se.

0:03:56.560 --> 0:03:59.920
<v Speaker 4>This is about them being able to unlock new AI

0:04:00.400 --> 0:04:04.360
<v Speaker 4>workflows and that really lets them tailor what they're doing

0:04:04.400 --> 0:04:06.240
<v Speaker 4>in complex media creation.

0:04:06.480 --> 0:04:08.240
<v Speaker 2>When you say their own use, couldn't they use it

0:04:08.280 --> 0:04:10.360
<v Speaker 2>on social media and say, okay, this is you know,

0:04:10.400 --> 0:04:12.200
<v Speaker 2>this is a way that we can connect with our

0:04:12.520 --> 0:04:15.400
<v Speaker 2>audience through Instagram or through TikTok or something. Creating one

0:04:15.400 --> 0:04:16.920
<v Speaker 2>of these videos and then it's out in the world.

0:04:17.800 --> 0:04:20.360
<v Speaker 4>Yeah, I'll give you an example of one that Paramount

0:04:20.400 --> 0:04:23.800
<v Speaker 4>plus died last year. They put you know, the if

0:04:23.839 --> 0:04:27.320
<v Speaker 4>campaign out there. That's a way to let fandom drive

0:04:28.000 --> 0:04:30.920
<v Speaker 4>part of the brand, but they're using a responsible model

0:04:30.960 --> 0:04:34.200
<v Speaker 4>behind it. Same thing with Gatorade. Gatorade has put out

0:04:34.240 --> 0:04:38.560
<v Speaker 4>an AI powered firefly model on their website, and that's

0:04:38.600 --> 0:04:40.880
<v Speaker 4>a way for them to be able to control the brand,

0:04:41.279 --> 0:04:43.240
<v Speaker 4>put it in the hands of fandom, because we all

0:04:43.279 --> 0:04:45.360
<v Speaker 4>want to be part of the narrative now. So that's

0:04:45.400 --> 0:04:49.040
<v Speaker 4>the responsible approach that they both those brands decided to take.

0:04:49.320 --> 0:04:51.120
<v Speaker 3>So put it out when they want, but they're also

0:04:51.200 --> 0:04:54.840
<v Speaker 3>creating kind of their own. It almost feels like private

0:04:55.160 --> 0:04:59.200
<v Speaker 3>set of images or IP as you said, that is

0:04:59.360 --> 0:05:02.040
<v Speaker 3>really for them to use and to control.

0:05:02.240 --> 0:05:03.040
<v Speaker 1>Correct.

0:05:03.720 --> 0:05:06.839
<v Speaker 4>Yes, I mean again, what you know, lots of these

0:05:06.839 --> 0:05:10.520
<v Speaker 4>companies have large IP that they have built over time

0:05:10.600 --> 0:05:13.719
<v Speaker 4>their beloved franchises that we know or brands that we know,

0:05:14.360 --> 0:05:16.080
<v Speaker 4>and they wanted to be in the driver's seat. So

0:05:16.120 --> 0:05:19.640
<v Speaker 4>they actually came to us and said, on this AI journey,

0:05:19.720 --> 0:05:22.440
<v Speaker 4>we trust Adobe. Adobe has been with us in marketing,

0:05:22.800 --> 0:05:24.839
<v Speaker 4>has been with us in creative We feel you have

0:05:24.920 --> 0:05:27.680
<v Speaker 4>a good AI approach to this and you're the one

0:05:27.680 --> 0:05:30.159
<v Speaker 4>that we want to partner with. So whether it's in

0:05:30.200 --> 0:05:33.600
<v Speaker 4>our products like Jen's Studio for performance marketing or photoshop,

0:05:33.720 --> 0:05:36.479
<v Speaker 4>like I said, we are part of the trusted workflow

0:05:36.520 --> 0:05:38.640
<v Speaker 4>that they're using and that's why they came to us.

0:05:38.720 --> 0:05:40.800
<v Speaker 3>All right, So then let's talk about your business and

0:05:40.800 --> 0:05:45.520
<v Speaker 3>what this does to drive usage above and beyond what

0:05:45.640 --> 0:05:47.320
<v Speaker 3>enterprises are already doing with you.

0:05:49.680 --> 0:05:52.720
<v Speaker 4>Sure, this is really just an evolution, a natural evolution

0:05:52.880 --> 0:05:55.280
<v Speaker 4>of our AI journey that started. Thank you for covering

0:05:55.279 --> 0:05:58.720
<v Speaker 4>the firefly image models in twenty twenty three. Businesses are

0:05:58.720 --> 0:06:01.480
<v Speaker 4>coming in US and saying, hey, we feel really good

0:06:01.520 --> 0:06:04.640
<v Speaker 4>about what we've been doing. The initial base models actually

0:06:04.680 --> 0:06:07.719
<v Speaker 4>didn't understand their IP. We consider that a future, not

0:06:07.760 --> 0:06:10.520
<v Speaker 4>a bug of those models, right, because we had licensed

0:06:10.520 --> 0:06:12.880
<v Speaker 4>to every piece that went in. And so as they've

0:06:13.160 --> 0:06:16.839
<v Speaker 4>moved up the customization spectrum, they are now asking for

0:06:16.920 --> 0:06:20.039
<v Speaker 4>world building and that is what Foundry as an AI

0:06:20.160 --> 0:06:24.039
<v Speaker 4>solution or Adobe AI Foundry offers them. And so we're

0:06:24.080 --> 0:06:27.479
<v Speaker 4>really pleased to have trust from the biggest fortune two

0:06:27.480 --> 0:06:28.599
<v Speaker 4>thousand brands out there.

0:06:29.839 --> 0:06:31.640
<v Speaker 3>One of the things that we wonder is what's the

0:06:31.640 --> 0:06:34.640
<v Speaker 3>model is trained by one of your customers or clients.

0:06:35.480 --> 0:06:38.320
<v Speaker 3>Do they need to keep paying to use it or

0:06:38.360 --> 0:06:41.040
<v Speaker 3>is this a one time consulting type charge or ongoing

0:06:41.040 --> 0:06:42.479
<v Speaker 3>subscription revenue for you?

0:06:43.640 --> 0:06:45.840
<v Speaker 4>Yeah, so there's always innovation. I think you see new

0:06:45.880 --> 0:06:48.560
<v Speaker 4>model releases are coming out all the time, and so

0:06:48.640 --> 0:06:52.160
<v Speaker 4>this is an ongoing relationship actually, you know, they've asked

0:06:52.240 --> 0:06:54.039
<v Speaker 4>us to be more like partners with them on this

0:06:54.120 --> 0:06:58.400
<v Speaker 4>long term journey. So there's evolutions of this both from image.

0:06:58.600 --> 0:07:01.159
<v Speaker 4>Maybe you're moving into video, we're moving into three D,

0:07:01.600 --> 0:07:04.279
<v Speaker 4>and of course as new architectures and models come out,

0:07:04.560 --> 0:07:06.039
<v Speaker 4>we'll continue that journey with them.

0:07:06.839 --> 0:07:08.279
<v Speaker 2>You know, we only have about a minute left, But

0:07:08.279 --> 0:07:09.960
<v Speaker 2>I got to be honest, Carol and I are a

0:07:09.960 --> 0:07:12.800
<v Speaker 2>little freaked out by this technology and the implications of

0:07:12.840 --> 0:07:14.800
<v Speaker 2>it a lot and just how good it's going to

0:07:14.800 --> 0:07:16.960
<v Speaker 2>get convince us that we shouldn't be freaked out.

0:07:18.640 --> 0:07:20.960
<v Speaker 4>I don't think you should be freaked out because at

0:07:21.000 --> 0:07:24.840
<v Speaker 4>the essence of what we believe is human creativity and

0:07:24.880 --> 0:07:27.160
<v Speaker 4>so tools like Photoshop. I'll just give you a quick

0:07:27.160 --> 0:07:30.520
<v Speaker 4>story when photoshop launched in the nineties, that was also

0:07:30.640 --> 0:07:33.360
<v Speaker 4>kind of a transformative thing, and it was not well embraced.

0:07:33.360 --> 0:07:35.600
<v Speaker 4>Now it's a verb to us. So these are new

0:07:35.600 --> 0:07:38.720
<v Speaker 4>tools at the intersection of creativity that all of us

0:07:38.800 --> 0:07:40.320
<v Speaker 4>are going to be able to figure out how to

0:07:40.360 --> 0:07:43.040
<v Speaker 4>unleash our own narrative in storytelling. But I'll read about

0:07:43.040 --> 0:07:44.520
<v Speaker 4>it as a friendly intern.

0:07:44.960 --> 0:07:46.720
<v Speaker 3>Take you out ten seconds fifties that are we get

0:07:46.800 --> 0:07:48.440
<v Speaker 3>a moment, so we're like, wait, is that real or

0:07:48.520 --> 0:07:49.760
<v Speaker 3>is that fake? Like are we going to have a

0:07:49.800 --> 0:07:52.400
<v Speaker 3>hard time as it gets better and better just very quickly.

0:07:53.240 --> 0:07:53.480
<v Speaker 1>Yeah.

0:07:53.520 --> 0:07:56.920
<v Speaker 4>So we have content authenticity, which is part of our provenance.

0:07:57.440 --> 0:08:00.360
<v Speaker 4>Every single piece of data, every single piece of NAVE

0:08:00.440 --> 0:08:03.960
<v Speaker 4>is stamped indefinitely with a forensic warmack, So I don't

0:08:03.960 --> 0:08:05.679
<v Speaker 4>think you'll be wondering from a Jobe.

0:08:05.880 --> 0:08:10.640
<v Speaker 3>So we'll have real Tim, real Carol, faint Carol, fake Tim.

0:08:10.880 --> 0:08:15.080
<v Speaker 3>Hannah really cool stuff, interesting Hannah Elsaka. She's vice president

0:08:15.120 --> 0:08:18.000
<v Speaker 3>of jen a I New Business Ventures at Adobe.