Vai al lettorePassa al contenuto principale
  • 13 minuti fa

Categoria

📺
TV
Trascrizione
00:00always. Meantime, AI voice startup, 11 Labs, talking about AI, announcing it has surpassed
00:06500 million dollars now in annual reoccurring revenue. The company also bringing on some new
00:10investors, including BlackRock, NVIDIA, and celebrities like Jamie Foxx and Eva Longoria.
00:15The milestones coming as artists like Taylor Swift move to protect their voices through
00:20trademarks and growing concerns over AI deepfakes. Joining us right now is 11 Labs CEO, Matti.
00:26Matti, I always mispronounce your last name, so I'm not even going to try because I know I'm going to
00:30screw it up and then you're going to laugh at me. It's great to see you, though. Help us understand
00:36where we really are in this technology journey. Obviously, from a financial perspective,
00:44you guys have been killing it. The question I have is sort of where do you think all of this
00:48goes?
00:48And for those who are uninitiated, this is a model that could take even the conversation we have
00:53right now and effectively generate, you know, extraordinarily a voice that's almost identical,
01:00if not more so.
01:03Andrew, Melissa, great to be here to talk about what's happening at 11 Labs and across our customers.
01:09So maybe as a quick introduction, quick context, we launched our first product in 2023 and now,
01:14three years later, just crossed a 500 million ARR milestone and are bringing an incredible set of
01:21investors on board to help us in that next phase of growth. And as a company, what we do, we
01:26combine
01:27research, product, and deployment to help businesses transform how they communicate. And that really
01:33started with building the transformational AI voice and audio models and now delivering them
01:38across the platform that helps power entirety of interactions across the company. So whether that's
01:44sales, support, business operations with AI agents, whether that's marketing and media with creative
01:50tooling, and maybe some of the examples of the work we do, we'll be helping Deutsche Telekom,
01:54Klarna better deploy those conversational agents, bringing the voice, bringing the knowledge of that
02:00organization, bringing the context of the user to really elevate how they can serve their customers
02:06all the way to NVIDIA and RAMP helping across the creative work. And to answer your question,
02:12Andrew, I think we are just at the beginning of the adoption of how the businesses communicate across.
02:17Matt, explain this though to me, just so I understand. You know, we talk about your company
02:22in the context of voice, and you guys have some extraordinary voice models. And then we often
02:28talk about enterprise companies that are trying to support telephone agents and other things.
02:34I'm thinking of a Brett Taylor's company, for example, and others. How do you, you guys are
02:41increasingly doing a lot of enterprise work. So where do we think of you relative to some of these
02:45other players?
02:48You know, I think the biggest piece that we see is AI becomes smarter. And one clear piece is that
02:56communication becomes the bottleneck. We can build intelligent system. But if your AI sounds robotic,
03:02if it interacts strangely, it just won't be trusted or widely adopted. So as a company, we take the two
03:08together. We solve the research as one of the few companies in the space building those foundational
03:13models for interaction, how you make it seamless, how you make it sound great, alongside exactly what
03:19you mentioned. How do you bring the knowledge of the organization, how you bring the context
03:23of the user, how you bring the deployment capacity to evaluate, monitor the agents in the production.
03:29I think the biggest difference that we took as a company is that we built a configurable open
03:34platform that can be used in support of some of the other companies, but can also be used in sales,
03:40in proactive outbound, in business operation, in training, in hiring. So maybe a good example of
03:46that is Deutsche Telekom. Deutsche Telekom started our work by helping on the marketing side, so creating
03:52a better materials for their audience. That then extended to customer support, where everybody can
03:58reach the T-Mobile network and have conversational agents that serve that better, all the way to now
04:03deploying the first of its kind agent inside of the network. So when you are calling through phone,
04:09as a T-Mobile subscriber, you can ask an agent to help you get additional information, all the way
04:16to real-time translate your conversation to another language. And that's what we are seeing across the
04:22last six, 12 months of growth, which is that enterprise adoption across those domains.
04:27Mani, what do you see as the moat around your business? The reason I ask is, it seems like almost
04:34every day, the next generation frontier model is able and capable of doing things we never thought
04:41was possible, and effectively entering new spaces.
04:47Yeah, the way we look at it, it's a combination of research, product, and ecosystem we create across.
04:53So research, my co-founder, my best friend, is one of the leading researchers in the space,
04:58of course, is able to create models that lead on benchmarks against the biggest AI labs out there,
05:03whether that started with text-to-speech, speech-to-text, or that interactive ability for agents.
05:08And we think that will compound over time as you refine the models for specific industries like
05:14healthcare, financial services, with combination of data, with combination of how to make them
05:19reliable in those settings. The product, here, you need to understand the workflow, you need to
05:24understand the problems of the customers. And to your earlier question, you really need to go deeper
05:28into each of the domains to build the platform that's very subtle. And the last piece is the
05:35ecosystem, and that's a little bit to why our investor represents Saturday vs. Set, where we are not
05:41only creating the product, but bringing that in combination between tech and creative space by
05:47creating an ecosystem where people can contribute voices, contribute templates for agents.
05:51Well, that's where I wanted to go with this. Let's just talk about the IP behind these voices,
05:54because the truth is people like Taylor Swift and others have brought lawsuits. Everyone's very
05:59worried about their IP. I should be worried about my own voice. And now I know you're getting into
06:05business with people like Jamie Foxx and Eva Liguria. So how is this going to work?
06:11As a company, we took a very specific approach where we want to work with the industry to disrupt some
06:18of
06:18the process and learn how the storytelling of the future can be made. That started all the way when
06:24we launched the first product. We created the first effectively voice marketplace where anyone can
06:29create their voice. We authenticate it. Then you can share it. And while the voice is being used,
06:34you can earn passively compensation as a result. Today, we have 20,000 voices on the platform that
06:39people can work and expanded this model to music and recently to agent templates where you can
06:44pre-configure agents so people can build on top of as well. And that's where we are bringing a lot
06:50of investors on board to be able to learn from them of how they see the future of storytelling,
06:54how they see the future of media, and bring those learnings into the platform. Maybe a great example
06:59of some of that work has been recent with Matthew McConaughey, where Matthew creates his newsletter
07:05that he narrates in English. But for the first time ever, we brought that over to Spanish,
07:10to Portuguese, where he can enjoy with his incredible voice, intonation, and emotion,
07:15that same newsletter with that new language set. And that's something we'll see increasingly
07:20across the companies we work with.
07:23Do you think long-term that this means that the incumbents, that this is going to benefit
07:30the incumbents, meaning somebody who's famous, currently famous, will have a much better shot of people
07:37who are going to come up in the future, because it's going to make it harder for people to come
07:42up
07:42in the future, because the incumbents are going to have a bigger share?
07:48There are two patterns that are happening. Of course, one is with the democratization of creative tools,
07:55everyone can get started without the same access of expressive equipment that was needed in the past.
08:02So I think that's the first pattern where you will see increasing amount of people just being able to break
08:06through
08:07that might have not had the resources before together. Now, of course, you actually now need the talent,
08:12you need the story, you need the emotions that you can expose, and that will be the differentiating part.
08:17There will be a second pattern, and to your question, where people that already build the audience
08:23will be able to expand their work in just completely new ways,
08:27whether that's breaking down language barrier and speaking other languages,
08:30whether that's shifting from static to interactive content,
08:33where in the past, that delivery just happened once,
08:36now you will be able to interact with those characters in a completely new setting,
08:39whether that's in the gaming environment, whether that's through a fun engagement campaign,
08:44all the way through to new educational capacities that were impossible before.
08:49Recently, we had a pleasure of working with Masterclass,
08:51where in Masterclass, previously, you had a static content delivery for educational lessons,
08:56now you can learn cooking by speaking with Gordon Ramsey,
08:59all the way through to learning negotiation by negotiating life with Chris Voss.
09:04And I think we'll see increasing amounts of that.
09:06Okay, final question. It's a security question.
09:09Do you and your family now have what they call a safe word because of these kind of models?
09:15Everybody's so scared that these models are so good that, you know,
09:18you could have your brother or sister or wife or kids call you up on the phone,
09:24or you think they're calling on the phone because the voice is so good,
09:27and you need to have a safe word that they don't otherwise know?
09:32Look, it's a great question.
09:33I think in general, we think you will see increasing amounts of AI content, AI interaction, AI agents in the
09:38future.
09:38All of us will have an AI agent that interacts on our behalf, but then she can call us.
09:42So, yes, we'll need a new safety systems in place.
09:46As a company, we do provide a lot of new safeguards, new abilities of how you can identify and authenticate
09:54that across,
09:55whether that's provenance and traceability, whether that's moderation,
09:58but also collaborating with the industry on new mechanisms like SYNFID or C2PA.
10:03To your pointed question, we do have a set of mechanisms with my close friends to make sure that as
10:12I'm a public,
10:13increasingly a public figure, they know how to reach me or when I'm reaching out in terms of need.
10:20And we are speaking with governments of how to make sure that the same mechanism exists for the way to
10:25space.
10:25And, Matty, thank you.
10:27Congratulations.
10:27Look forward to talking again soon.
10:29And talking to the real version, too.
10:31I enjoy talking to the real version of you.