WEBVTT - Amazon's Anthropic Deal and Hollywood Screenwriters' Deal 0:00:01.440 --> 0:00:05.080 From Marhart where Innovation of Money and Power Collie in 0:00:05.200 --> 0:00:06.720 Silicon Valley, NBN. 0:00:07.040 --> 0:00:11.080 This is Bloomberg Technology with Caroline Hyde and Ed Ludlow. 0:00:25.200 --> 0:00:27.760 I'm Caroline Hand of Bloomberg's World headquarters in New York, 0:00:28.360 --> 0:00:30.440 and I May Lovelow back in San Francisco. 0:00:30.640 --> 0:00:32.879 This is Bloomberg Technology coming up. 0:00:32.960 --> 0:00:35.920 Amazon pushing further into the world of generative AI with 0:00:36.000 --> 0:00:36.560 an up. 0:00:36.440 --> 0:00:38.680 To four billion dollar deal with Anthropic. 0:00:38.920 --> 0:00:40.960 We're going to be sitting down with Amazon Web Services 0:00:41.000 --> 0:00:44.200 CEO Adam Slipski to discuss. 0:00:44.080 --> 0:00:48.280 And Hollywood screenwriters strike a tentative deal with studios to 0:00:48.440 --> 0:00:51.559 end their month's long strike. Rereate down everything you need 0:00:51.600 --> 0:00:52.640 to know about the agreement. 0:00:52.760 --> 0:00:54.760 Plus Apple's cheapest iPhone fifteen. 0:00:54.840 --> 0:00:57.000 Well, it's winning over buyers, with wait times for the 0:00:57.040 --> 0:01:00.000 device almost doubling compared to this time last year. 0:01:01.760 --> 0:01:04.320 Yep, let's get straight to our top short story. Amazon 0:01:04.360 --> 0:01:07.360 will invest up to four billion dollars in the AI 0:01:07.520 --> 0:01:11.880 startup Andthropic. It is a complicated deal where Anthropic will 0:01:11.920 --> 0:01:16.160 move its software to AWS cloud, but also the training 0:01:16.520 --> 0:01:20.440 of its future models to Amazon's proprietary silicon. We are 0:01:20.480 --> 0:01:25.120 talking about the AI accelerators Trainium and Inferentia. This is 0:01:25.200 --> 0:01:27.720 kind of an interesting move for Amazon that's done so 0:01:27.880 --> 0:01:32.319 much work internally on generative AI tools, but now it 0:01:32.360 --> 0:01:35.040 looks to a third party. Of course, the clawed large 0:01:35.120 --> 0:01:39.039 language model is one with billions of parameters. It is 0:01:39.040 --> 0:01:42.679 one that was already available within Bedrock, the platform that 0:01:42.800 --> 0:01:46.520 Amazon through AWS offers to build your own llms or 0:01:46.560 --> 0:01:51.040 take advantage of pre built skeletons of llms. There are 0:01:51.120 --> 0:01:54.320 many questions to be posed about the strategy here from 0:01:54.320 --> 0:01:58.680 Amazon and AWS with regards to generative AI and what's 0:01:58.720 --> 0:02:01.120 going on with their propriety to take We want to 0:02:01.120 --> 0:02:05.360 welcome our Bloomberg television and radio audiences worldwide joining us 0:02:05.400 --> 0:02:09.720 now Amazon Web Services CEO Adam Selipski, and Adam welcome 0:02:09.760 --> 0:02:11.840 to Bloomberg Technology. I want to start with some of 0:02:11.840 --> 0:02:16.519 the mechanics of this deal. Does Anthropic still pay Amazon 0:02:16.600 --> 0:02:20.840 to use AWS Cloud or is it structured such that 0:02:21.320 --> 0:02:23.799 the investment that you make in Anthropic is in the 0:02:23.800 --> 0:02:27.000 form of cash and credits for AWS Cloud? 0:02:28.480 --> 0:02:30.400 No, good morning, Thanks for having me. 0:02:30.440 --> 0:02:36.079 We're very excited for this expanded relationship with Anthropic, and 0:02:36.760 --> 0:02:40.839 the investment is a financial investment, as you say, and 0:02:41.360 --> 0:02:47.000 In addition, Anthropic will be training future versions of its 0:02:47.000 --> 0:02:52.000 models and running its models on AWS using our training 0:02:52.520 --> 0:02:57.000 chips and inferential chips. Those models will be guaranteed to 0:02:57.000 --> 0:03:00.200 be available for years to come. In our amaz On 0:03:00.280 --> 0:03:04.120 Bedrock managed service for llms, which provides the very wide 0:03:04.200 --> 0:03:08.840 choice of models and AWS, customers will actually receive early 0:03:08.960 --> 0:03:12.720 access to key features in Anthropics models in the future, 0:03:12.800 --> 0:03:16.960 such as fine tuning and customization of models. In addition, 0:03:17.960 --> 0:03:22.640 Anthropics are very talented technical teams and we anticipate working 0:03:22.680 --> 0:03:26.079 closely with them to actually improve future versions of our 0:03:26.120 --> 0:03:30.239 training in Inferentia at chips. So there are a lot 0:03:30.240 --> 0:03:33.000 of different benefits for our joint end customers from this 0:03:33.080 --> 0:03:36.040 relationship and we're very excited to be leaders in this together. 0:03:37.240 --> 0:03:39.920 Adam, there's a lot of emphasis on moving a maker 0:03:39.960 --> 0:03:43.160 of foundation models at that scale once your proprietary silicon. 0:03:43.600 --> 0:03:48.760 How quickly will Anthropics start running AI workloads on Trainium 0:03:48.760 --> 0:03:49.520 and Inferentia. 0:03:50.680 --> 0:03:54.000 Well, we've been working with Anthropic, they've been a customers 0:03:54.280 --> 0:03:57.360 of ours since I think they're founding over a couple 0:03:57.400 --> 0:04:00.200 of years ago, and so they use a variety of 0:04:00.240 --> 0:04:04.360 different technologies for a variety of different workloads on AWS, 0:04:07.600 --> 0:04:11.760 that they'll be using GPUs on AWS and will also 0:04:11.800 --> 0:04:15.200 be using large quantities of TRAININGUM and inferentia. 0:04:15.320 --> 0:04:17.280 So I think everything's going to move very. 0:04:17.240 --> 0:04:21.040 Quickly and it'll all be a mix of technologies depending 0:04:21.040 --> 0:04:21.960 on their needs at the time. 0:04:23.240 --> 0:04:26.800 Adam, what's the mood like within Amazon and AWS this morning? 0:04:26.800 --> 0:04:30.240 There are lots of talented engineers that have been working 0:04:30.279 --> 0:04:34.520 on large language models generative AI tools internally, and now 0:04:34.560 --> 0:04:37.840 you're turning to a third party who's highly regarded as 0:04:37.839 --> 0:04:40.240 a leader in building foundation models. 0:04:42.200 --> 0:04:43.239 The mood here is great. 0:04:43.920 --> 0:04:46.320 We are a company of inventors who we love to build, 0:04:46.800 --> 0:04:48.800 and there's never been a better time to be a 0:04:48.800 --> 0:04:54.080 builder at AWS than right now. And as I mentioned before, 0:04:55.120 --> 0:04:57.640 a big part of our strategy in AI and generative 0:04:57.640 --> 0:05:01.360 AI specifically, it's all about customer choice and there's not 0:05:01.440 --> 0:05:05.320 going to be any one solution that works for all 0:05:05.320 --> 0:05:08.440 customers for all use cases. And Thropic has done an 0:05:08.440 --> 0:05:12.400 amazing job. They're clearly a leader in this space, and 0:05:12.480 --> 0:05:15.520 it's really important for customers that we continue to generate 0:05:15.560 --> 0:05:19.919 new capabilities together at the same time. Really one of 0:05:19.920 --> 0:05:24.040 the hallmarks of our Amazon Bedrock Managed service for generative 0:05:24.040 --> 0:05:27.599 AI is choice, and so Amazon is going to continue 0:05:27.600 --> 0:05:29.840 to build its own Titan models, which are going to 0:05:29.839 --> 0:05:35.640 be available later this year. Obviously Onthropics models are prominent 0:05:35.680 --> 0:05:38.760 in bedrock, and we will have models from other leading 0:05:38.800 --> 0:05:41.839 providers as well as we have today. So it's still 0:05:41.839 --> 0:05:45.159 an amazing time to build here at Amazon. We think 0:05:45.160 --> 0:05:46.680 our models are going to be great as well, and 0:05:46.720 --> 0:05:49.559 it's about customers choosing the right tool for the job. 0:05:50.320 --> 0:05:53.000 Talking about choice, and I just want to re welcome 0:05:53.040 --> 0:05:56.880 our TV and radio audiances with Adam Sleipski. What's so 0:05:57.040 --> 0:06:01.000 notable is that well Anthropic took a change one hundred 0:06:01.000 --> 0:06:05.360 million dollars worth from Google already, and I'm interested as 0:06:05.400 --> 0:06:07.960 to how you feel that is perhaps a concern for 0:06:08.040 --> 0:06:11.800 you or not the relationship that athropag already has with 0:06:12.279 --> 0:06:13.479 a previous cloud provider. 0:06:14.760 --> 0:06:18.240 Now, we feel great about the relationship with Anthropic. It's 0:06:18.279 --> 0:06:21.159 been a good relationship and I think today's announcement just 0:06:21.240 --> 0:06:27.680 makes it a deeper and longer term. Anthropic will use 0:06:27.720 --> 0:06:31.600 AWS as its primary cloud provider for mission critical workloads, 0:06:31.600 --> 0:06:38.000 including building foundational foundation models and doing AI safety research, 0:06:38.279 --> 0:06:41.719 and will run the majority of its workloads on AWS. 0:06:41.760 --> 0:06:44.599 So we feel great about being able to provide the 0:06:44.640 --> 0:06:48.080 capacity and the expertise, and of course the security, the 0:06:48.200 --> 0:06:51.919 enterprise grade security that is so important to AWS customers. 0:06:52.080 --> 0:06:54.479 And we also feel great about working with Anthropic to 0:06:54.520 --> 0:06:58.880 make sure that our trainum and inferential technology, our chips 0:06:58.920 --> 0:07:02.320 are as as cutting edge as possible going forward for 0:07:02.400 --> 0:07:03.000 years to come. 0:07:03.680 --> 0:07:06.080 I'm interested in drilling down sort of why ED was 0:07:06.120 --> 0:07:09.000 going about the feeling internally right now, because I look 0:07:09.000 --> 0:07:11.680 at some of the analyst reaction to this. Adam and Webbush, 0:07:11.680 --> 0:07:15.320 for example, they say this signals a newfound urgency in 0:07:15.400 --> 0:07:18.920 Amazon's strategy to further integrate generative AI among your AW. 0:07:18.880 --> 0:07:22.640 Suite of services. That urgency was there a lack. 0:07:22.480 --> 0:07:26.600 Of understanding or indeed a reality that Amazon was behind 0:07:26.640 --> 0:07:28.640 the curve here a little bit when it came to 0:07:29.440 --> 0:07:32.040 the integration of generative AI. Because we've been looking at 0:07:32.080 --> 0:07:33.960 open Ai and Microsoft for a while now. 0:07:34.880 --> 0:07:38.200 We've been saying for many, many months, Carolyn, that we 0:07:38.240 --> 0:07:40.960 are fully urgent. We have a strategy that we really love. 0:07:41.320 --> 0:07:44.000 It is different than some other cloud provider strategies. It's 0:07:44.080 --> 0:07:50.480 true we have a strategy of providing absolutely uncompromising security, 0:07:50.840 --> 0:07:53.080 which I don't think is true for our cloud providers. 0:07:53.280 --> 0:07:56.400 We have a strategy of providing customers the choices to 0:07:56.520 --> 0:08:00.800 use whatever's best for their job at hand. So Andthropic 0:08:00.920 --> 0:08:03.560 is going to be an amazing set of models for many, 0:08:03.600 --> 0:08:08.240 many use cases. And Amazon is fully invested in building 0:08:08.240 --> 0:08:10.960 its own Titan models, which I think will be really 0:08:11.040 --> 0:08:14.640 useful for other customers and other circumstances, And of course 0:08:14.640 --> 0:08:19.960 are other model provider partners through Bedrock. So I really 0:08:20.400 --> 0:08:22.440 think it's an ill founded premise that there's been some 0:08:22.560 --> 0:08:25.520 change in urgency. We're fully urgent here on generative AI 0:08:26.040 --> 0:08:28.360 for one reason and one reason alone. It's because our 0:08:28.400 --> 0:08:32.320 customers need us to have great generative AI capabilities. So 0:08:32.400 --> 0:08:35.160 many of them have their data platforms on AWS, and 0:08:35.240 --> 0:08:38.640 if you got your data here, you really want to 0:08:38.640 --> 0:08:41.760 have your generative AI and all the powerful capabilities that 0:08:41.800 --> 0:08:44.680 you need from those capabilities in the same place. And 0:08:44.679 --> 0:08:47.720 so we have been, are and will continue to be 0:08:47.880 --> 0:08:49.720 very motivated to deliver for customers. 0:08:50.559 --> 0:08:52.760 Adam, what does this mean for the kind of rampop 0:08:52.840 --> 0:08:56.000 or path forward for TRAININGUM and inferentia. You've put a 0:08:56.040 --> 0:08:59.600 lot of emphasis that anthrop It brings you a maker 0:08:59.679 --> 0:09:03.560 of creator foundation models at scale. We now need to 0:09:03.679 --> 0:09:08.400 ramp up I guess your third party manufacturing relationships to say, okay, 0:09:08.440 --> 0:09:11.960 let's get more trainium on more inferential online to support 0:09:12.000 --> 0:09:12.760 the workloads. 0:09:13.880 --> 0:09:17.160 Well, it's absolutely true that there is a huge demand 0:09:17.760 --> 0:09:22.040 for all of the different ships with which people do 0:09:22.559 --> 0:09:26.600 a generative AI workloads, and so we absolutely have already 0:09:26.640 --> 0:09:31.160 been ramping up our training and inferential supply chain and 0:09:31.440 --> 0:09:34.400 ramping up the supply that we can create as quickly 0:09:34.440 --> 0:09:38.840 as possible. And yes, Anthropic will have access to very 0:09:38.840 --> 0:09:43.160 significant quantities of compute which we'll have trainum and inferentia 0:09:43.160 --> 0:09:46.920 in them. So yes, that's one of many reasons why 0:09:46.960 --> 0:09:49.880 we continue to ramp up and to provide a very 0:09:49.960 --> 0:09:54.120 robust AWS controlled supply chain for AI chips. 0:09:54.520 --> 0:09:57.959 And is that where the revenue boost comes, Adam, because 0:09:58.000 --> 0:09:58.520 we're looking at. 0:09:58.440 --> 0:10:01.200 The share price reaction is high on the day. When 0:10:01.240 --> 0:10:03.080 does this all start to really. 0:10:02.920 --> 0:10:06.319 Drive adoption money and the bottom line for Amazon? 0:10:07.640 --> 0:10:09.960 Well, I think that AI in general. 0:10:10.080 --> 0:10:13.600 Look, AWS has had machine learning services since at least 0:10:13.600 --> 0:10:18.120 twenty seventeen when we released our sage Maker machine learning service, 0:10:18.160 --> 0:10:21.040 which has over one hundred thousand AWS customers on it. 0:10:21.080 --> 0:10:23.600 So we've been doing machine learning for a long time 0:10:23.720 --> 0:10:27.720 inside of AWS, and obviously more recently have had a 0:10:27.760 --> 0:10:32.120 significant number of Generative AI customers, and we will certainly 0:10:32.679 --> 0:10:35.959 continue to ramp up anticipate quite steeply. We have many 0:10:36.040 --> 0:10:40.600 sources of growth inside of AWS where a scaled and 0:10:40.600 --> 0:10:45.320 relatively sizable business at this point, and customers are running 0:10:45.320 --> 0:10:49.920 their data platforms on AWS. They are building out more 0:10:49.960 --> 0:10:54.000 and more applications for things like supply chain and contact 0:10:54.040 --> 0:10:56.920 center management on AWS. I've still a whole lot of 0:10:56.960 --> 0:11:00.000 storage and compute and database workloads ramping on our ADA, 0:11:00.600 --> 0:11:04.440 So we have many sources of growth anticipate. But there's 0:11:04.480 --> 0:11:07.520 absolutely no doubt that generative AI looks like it's going 0:11:07.559 --> 0:11:10.679 to be an explosive additional source of growth in the 0:11:10.760 --> 0:11:11.360 years ahead. 0:11:12.840 --> 0:11:15.120 Adam, we put a lot of emphasis on the up 0:11:15.160 --> 0:11:18.680 to four billion dollars, and you know, I understand and 0:11:18.760 --> 0:11:22.479 thank you for explaining how the relationship will work in practice. 0:11:22.760 --> 0:11:24.840 If I put to you this is an example of 0:11:24.880 --> 0:11:28.640 Amazon or AWS basically paying a leader in the field 0:11:28.640 --> 0:11:32.560 of AI, handing over cash to allow to make them 0:11:32.720 --> 0:11:35.920 use trainingum and inferentia, how would you respond to that 0:11:35.960 --> 0:11:38.200 and explain to me how you bring new customers on 0:11:38.280 --> 0:11:41.640 board who are really interested in the AI accelerators that 0:11:41.679 --> 0:11:44.800 you have built without having to invest in them. 0:11:45.880 --> 0:11:46.880 Is a sort of backup. 0:11:47.960 --> 0:11:52.400 Sure well, the I think the really big news today 0:11:52.640 --> 0:11:58.520 is the new expanded relationship between and Propic and Amazon, 0:11:59.320 --> 0:12:03.839 in which they will have access to really large quantities 0:12:03.920 --> 0:12:07.840 of trainingmen and fr at chips. Customers will have access 0:12:07.840 --> 0:12:11.600 to those models, including early access to critical features through 0:12:11.640 --> 0:12:15.880 Amazon Bedrock and Amazon will get to aws, We'll get 0:12:15.880 --> 0:12:20.040 to work with Anthropic to ensure that we optimize our 0:12:20.080 --> 0:12:22.160 trainingment in Ferentia technology going forward. 0:12:22.360 --> 0:12:25.480 That's the benefit for customers. And yes, as part of this. 0:12:25.800 --> 0:12:29.120 We're pleased to be making an additional an initial investment 0:12:29.160 --> 0:12:32.439 of one point twenty five billion dollars into Anthropic. It's 0:12:32.440 --> 0:12:36.079 a financial investment and that could go up as high 0:12:36.080 --> 0:12:39.559 as four billions, as you said, over time. But it's 0:12:39.600 --> 0:12:41.840 really driven around customer value and what this is going 0:12:41.880 --> 0:12:46.000 to mean to customers who are very, very determined as 0:12:46.040 --> 0:12:49.960 they should be, to figure out generative AI strategies. We 0:12:50.040 --> 0:12:53.600 already are working in depth with customers, as is Anthropic 0:12:54.320 --> 0:12:55.160 on forming those. 0:12:55.000 --> 0:12:57.040 Strategies and actually moving to execution. 0:12:57.360 --> 0:13:01.000 We have a lot of great customers from Lonely Planet, Nexus, Lexis, 0:13:01.840 --> 0:13:04.880 and a number of others who are actually moving to 0:13:04.920 --> 0:13:09.840 production with Generative AI on AWS and Anthropic And in addition, 0:13:09.920 --> 0:13:13.480 as you alluded to, we'll be working with all of 0:13:13.520 --> 0:13:16.760 the partners that our customers want to do business with. 0:13:16.880 --> 0:13:19.480 If it's an important partner to our customers, it's going 0:13:19.520 --> 0:13:21.280 to be an important partner to us as well. 0:13:22.760 --> 0:13:27.760 Amazon Web Services CEO Adam Slipski, thank you, Thank you. 0:13:30.160 --> 0:13:33.520 Hollywood writers and studios are reaching at tentative deal with 0:13:33.600 --> 0:13:35.720 on a sund of the weekend to end strikes which 0:13:35.760 --> 0:13:38.800 began back in May A. Stick into this with someone 0:13:38.840 --> 0:13:42.120 who's been on it nonstoplum Merg's Luca Sure, and can 0:13:42.160 --> 0:13:44.360 you just go to the intricacies of what we think 0:13:44.360 --> 0:13:45.160 a deal has been. 0:13:45.040 --> 0:13:45.720 Struck on here. 0:13:46.600 --> 0:13:49.319 So the biggest issues are the ones that they hadn't 0:13:49.360 --> 0:13:52.640 already reached some kind of deal on are related to 0:13:52.679 --> 0:13:55.839 minimum staffing, so having a minimum number of writers who 0:13:55.840 --> 0:13:59.240 are hired to work on a TV program, and then 0:13:59.400 --> 0:14:03.280 data trans parency, slash payment and success, so studios and 0:14:03.320 --> 0:14:07.120 streaming services being more open about how many people are 0:14:07.160 --> 0:14:11.600 watching their shows and then paying writers' bonuses. Basically if 0:14:11.640 --> 0:14:14.040 they perform up to a certain level, and then some 0:14:14.200 --> 0:14:17.640 agreement around artificial intelligence and what studios can and can't 0:14:17.640 --> 0:14:19.960 do with the scripts they own. I have some sense 0:14:20.000 --> 0:14:22.680 of the particulars on each of those points, but they 0:14:22.720 --> 0:14:25.480 haven't released the full details of the proposal just yet. 0:14:26.400 --> 0:14:29.200 You know, Lucas, the state of production is still worth considering. 0:14:29.280 --> 0:14:33.160 Right The actors strike is ongoing and talks are ongoing there. 0:14:33.640 --> 0:14:36.120 But the net result appears to be that, you know, 0:14:36.280 --> 0:14:39.640 late night talk shows that don't require actors, they might 0:14:39.720 --> 0:14:41.080 start getting underway soon. 0:14:42.360 --> 0:14:47.080 Yeah, the broader guild still needs to ratify this agreement. 0:14:47.120 --> 0:14:50.000 If sort of the negotiating committees for both sides that 0:14:50.040 --> 0:14:52.440 have agreed to this, there may be a condition of 0:14:52.480 --> 0:14:55.720 this where daytime talk shows late night talk shows are 0:14:55.760 --> 0:14:58.400 able to go back sooner than it's ratified. I believe 0:14:58.440 --> 0:15:00.280 that happened with the two thousand and two thousand, two 0:15:00.280 --> 0:15:03.120 thousand and seven, two thousand and eight writers strike. But 0:15:03.320 --> 0:15:06.000 either way, people expect the guilt to ratify and in 0:15:06.040 --> 0:15:08.160 the next couple of weeks we should see those shows 0:15:08.200 --> 0:15:10.760 get up and running. You're right that production on a 0:15:10.760 --> 0:15:14.400 lot of film and television can't happen just yet, but 0:15:14.520 --> 0:15:17.560 they could begin pre production, They could start writers' rooms 0:15:17.560 --> 0:15:19.120 for some of those shows you know they had. There 0:15:19.120 --> 0:15:20.920 were shows that were supposed to be on New in 0:15:20.960 --> 0:15:24.000 the fall in September that hadn't even started writing scripts yet. 0:15:24.000 --> 0:15:25.440 So that's something that could begin now. 0:15:26.040 --> 0:15:29.240 For many This is a relief for the writers who've 0:15:29.280 --> 0:15:32.240 been unable to work, but also the entire ecosystem that's 0:15:32.240 --> 0:15:32.960 built around them. 0:15:33.040 --> 0:15:34.720 Lucas and I'm interested. 0:15:34.400 --> 0:15:36.440 As to who are the key players, the key figures 0:15:36.480 --> 0:15:37.920 that got in the room to try and. 0:15:37.920 --> 0:15:38.600 Push this forward. 0:15:39.720 --> 0:15:42.800 Well, for most of the negotiation you had or the 0:15:43.160 --> 0:15:46.000 negotiating committee on the writer's side, and then the top 0:15:46.080 --> 0:15:48.560 labor lawyers on the side of the studios as well 0:15:48.560 --> 0:15:51.280 as they're representative with this body called the AMPTP, this 0:15:51.280 --> 0:15:55.560 woman Carol Lombardini. But you also saw in recent weeks 0:15:55.560 --> 0:15:58.320 and especially over the last week, the CEOs of some 0:15:58.360 --> 0:16:01.240 of the biggest media companies get person involved. That included 0:16:01.560 --> 0:16:05.200 Walt Disney CEO bar Biger, Netflix co CEO Ted Sorrandos, 0:16:05.240 --> 0:16:08.880 Warner Brothers, Discovery CEO David Saslav, and Donald Langley, who's 0:16:08.920 --> 0:16:12.160 the chief content officer for NBCUniversal. They were personally involved 0:16:12.160 --> 0:16:15.600 in the negotiations for much of last week, and are 0:16:16.360 --> 0:16:19.560 you know they're not necessarily the heroes of this situation, 0:16:19.720 --> 0:16:23.880 but their involvement did signal the seriousness with which the 0:16:23.920 --> 0:16:26.560 studios were approaching the negotiation and put something of a 0:16:27.040 --> 0:16:29.240 timetable on it. 0:16:29.320 --> 0:16:32.000 All right, Bloomberg's Lucas Shaw reporting through the weekend and 0:16:32.040 --> 0:16:35.360 here on Bloomba's technology, Thank you and a programming alert. 0:16:35.440 --> 0:16:38.720 In about two weeks, Lucas will be hosting Bloomberg's screen 0:16:38.800 --> 0:16:41.920 Time in La to cover the collision of Hollywood and 0:16:41.960 --> 0:16:46.440 Silicon Valleys, bringing together moguls, celebrities, and entrepreneurs already defining 0:16:46.720 --> 0:16:49.080 the next phase of pop culture. And if you're lucky, 0:16:49.240 --> 0:16:51.800 your two favorite Bloomberg technology anchors as well. 0:16:52.040 --> 0:16:52.880 This is Bloomberg. 0:17:01.840 --> 0:17:04.120 We started on General to AI. Let's go there again 0:17:04.200 --> 0:17:06.760 because the company forethought it's trying to take on some 0:17:06.840 --> 0:17:09.840 key competitors like Salesforce, for example, in the customer service 0:17:09.880 --> 0:17:13.080 department by launching auto Flows. It's a customer support tool 0:17:13.119 --> 0:17:13.840 using simple. 0:17:13.680 --> 0:17:16.320 Natural language prompts rather than manual. 0:17:15.960 --> 0:17:19.320 Workflows ticket based systems. The whole system is powered by 0:17:19.359 --> 0:17:22.359 Support GPT. That's four thoughts generative AI model, which is 0:17:22.359 --> 0:17:25.440 in itself is powered by open AI models. Let's bring 0:17:25.480 --> 0:17:28.359 in four thoughts CEO Deon Nicholas now for more and 0:17:28.440 --> 0:17:32.040 there's still and it's really interesting that you're really trying 0:17:32.080 --> 0:17:35.920 to drive home customer service here in a more natural, 0:17:36.119 --> 0:17:38.520 easier manner. Can you just say how much at the 0:17:38.560 --> 0:17:40.879 moment this is held back by lack of AI, how 0:17:40.960 --> 0:17:42.959 much you actually think it will change the game and 0:17:43.040 --> 0:17:44.919 take an individual out of the equation? 0:17:46.080 --> 0:17:49.240 Absolutely, Caroline, thanks for having me. And so when we 0:17:49.280 --> 0:17:53.159 think about the state of the art in AI right now, 0:17:53.400 --> 0:18:00.000 it's been hamstrung by manual workflows, decision trees, rules, keywords. 0:18:00.040 --> 0:18:02.760 And that's why every single chatbot we've ever interacted with 0:18:03.080 --> 0:18:05.920 we get that age old experience of hey, I need 0:18:05.960 --> 0:18:06.640 to talk to an agent. 0:18:06.640 --> 0:18:07.160 I don't really. 0:18:07.119 --> 0:18:09.480 Understand what's going on here, right, And so I think 0:18:09.520 --> 0:18:12.000 this is actually the future for customer. 0:18:11.680 --> 0:18:13.600 Support, not fully intererative. 0:18:13.680 --> 0:18:16.480 Yet you said within some of the discussion points it 0:18:16.520 --> 0:18:19.160 requires you still perhaps through the old additional task manually. 0:18:19.200 --> 0:18:21.240 But when do you think we might. 0:18:21.160 --> 0:18:23.119 Never have to have that age old I need to 0:18:23.119 --> 0:18:23.879 speak to an agent? 0:18:23.920 --> 0:18:24.120 Now? 0:18:26.200 --> 0:18:28.840 I think we're very close, right, with auto flows technology, 0:18:28.840 --> 0:18:32.439 you simply specify the goal or the prompt and you 0:18:32.520 --> 0:18:34.720 let AI do the rest. And this has become really 0:18:34.760 --> 0:18:39.119 really powerful where we've already seen this technology live in 0:18:39.160 --> 0:18:42.640 some of our private beta customers. We're launching or renouncing 0:18:42.640 --> 0:18:46.160 an open beata this past week, we've seen stats from 0:18:46.160 --> 0:18:49.000 some of our customers where the customer satisfaction scores have 0:18:49.040 --> 0:18:51.640 gone up by twenty seven percent. And we're already serving 0:18:51.640 --> 0:18:54.560 thousands of conversations on this, So we're already seeing the 0:18:54.640 --> 0:18:56.760 benefits of auto flows and I think the future is 0:18:56.760 --> 0:18:57.639 closer than we think. 0:18:58.520 --> 0:19:01.800 There are many audience members of limbot technology that will 0:19:01.800 --> 0:19:04.919 be used to the ticket based system. Some participants in 0:19:04.960 --> 0:19:08.160 the show also used to it give us one case study, 0:19:08.800 --> 0:19:11.680 a simple one that your technology will replace. 0:19:13.400 --> 0:19:16.320 Yeah, I think a lot of the ticketing system. So 0:19:16.359 --> 0:19:18.880 if you look at you know, Salesforce and folks like that, 0:19:20.040 --> 0:19:24.240 they've all been talking about this autonomous AI future. You've 0:19:24.280 --> 0:19:28.480 seen at the Dreamforce announcement, Einstein Copilot being announced that 0:19:28.520 --> 0:19:32.000 it's coming sometime in the future, sometime next year. We're 0:19:32.040 --> 0:19:36.520 already seeing use cases like Upwork, who've leveraged auto flows 0:19:36.560 --> 0:19:40.360 for a lot for a few of their workflows in 0:19:40.440 --> 0:19:45.200 their customer support center, and they've seen improvements in customer 0:19:45.200 --> 0:19:48.359 satisfaction and deflection rate by leveraging auto flows instead of 0:19:48.400 --> 0:19:50.520 traditional manual workflows. 0:19:50.520 --> 0:19:54.120 In some cases, support GBT is built on GPT three 0:19:54.119 --> 0:19:57.480 point five, GPT four, GPT four. So what's that been 0:19:57.600 --> 0:19:59.640 like in the few weeks and months into a last 0:19:59.640 --> 0:20:02.600 on the show building out the technology with that underpinning. 0:20:02.960 --> 0:20:06.120 Yeah, it's been it's been really fun. At forethought, it's 0:20:06.160 --> 0:20:07.720 kind of in the name. We've always been about bringing 0:20:07.720 --> 0:20:10.959 the future to now right, and so we've often leveraged 0:20:10.960 --> 0:20:13.439 our own technologies, our own models, and we also leverage 0:20:13.440 --> 0:20:15.840 the latest and greatest, and so it started with GPT 0:20:15.920 --> 0:20:17.280 three point five. I think that was it when I 0:20:17.320 --> 0:20:22.160 was last on this show, and we've focused on innovating. 0:20:22.240 --> 0:20:25.399 It's not just about large language models, which are in 0:20:25.440 --> 0:20:27.480 many ways new for people, but for us, we've seen 0:20:27.560 --> 0:20:29.639 large language models over the past few years, and we 0:20:29.640 --> 0:20:32.960 think the future is about autonomous AI or agentic AI, 0:20:33.200 --> 0:20:35.440 and so leveraging the latest and greatest technologies not just 0:20:35.480 --> 0:20:38.320 for language, but for action that can and AI that 0:20:38.359 --> 0:20:40.000 can go and take action on your behalf and solve 0:20:40.040 --> 0:20:42.320 customer problems has been huge. 0:20:42.480 --> 0:20:45.159 Thirty seconds, Deon, is this going to mean less jobs? 0:20:47.840 --> 0:20:51.120 I think long term, I'm extremely bullish that AI will 0:20:51.160 --> 0:20:55.160 actually create jobs for people. We saw the same in 0:20:55.440 --> 0:20:57.359 the Industrial Revolution, we saw the same in the move 0:20:57.440 --> 0:20:59.959 to the Internet. I think a lot of technologies are 0:21:00.040 --> 0:21:02.000 a lot of industries are going to be changed, and 0:21:02.160 --> 0:21:03.200 radically more jobs. 0:21:02.960 --> 0:21:04.160 Are going to be created over time. 0:21:04.600 --> 0:21:07.439 All right, four CEO d On Nicholas, a founder carrier 0:21:07.440 --> 0:21:09.960 that's building right here and was on the show earlier 0:21:09.960 --> 0:21:12.280 in the year. It's interesting to track the progress that 0:21:12.320 --> 0:21:14.120 he's made since then. Thank you very much. 0:21:21.760 --> 0:21:24.000 Welcome back to Rumog Technology. I'm Caroline Hide in New. 0:21:24.000 --> 0:21:27.000 York and I'm Ed Ludlow here in San Francisco. Okay, 0:21:27.040 --> 0:21:31.200 for today's going viral, we're focusing on football and Taylor 0:21:31.320 --> 0:21:34.680 Swift showing up to the Kansas City Chiefs game sitting 0:21:34.720 --> 0:21:38.280 in Travis Kelsey's box with his mother Donna. 0:21:38.320 --> 0:21:39.480 You see her there in Kara. I don't know if 0:21:39.480 --> 0:21:40.960 you saw this NFL on Fox. 0:21:41.000 --> 0:21:44.800 Handle on x posts this video of her celebrating that 0:21:44.840 --> 0:21:48.679 third quarter touchdown by Kelsey, and the numbers on that 0:21:48.800 --> 0:21:50.480 video astonishing. 0:21:51.119 --> 0:21:52.679 I mean, why does anyone know what the numbers were 0:21:52.680 --> 0:21:53.120 in the score. 0:21:53.280 --> 0:21:55.560 I feel like that none of that could actually focused 0:21:55.600 --> 0:21:57.000 on in the game. I think it was what ten 0:21:57.080 --> 0:22:00.800 forty one in the end, absolutely trounced the Bears. But ultimately, 0:22:00.880 --> 0:22:02.760 I mean, she said in what one of her songs 0:22:02.800 --> 0:22:04.760 arefore I think it's gold Rush that she liked the Eagles, 0:22:04.760 --> 0:22:07.320 which I think is where his brother plays. But all 0:22:07.359 --> 0:22:09.119