Rewriting the Enterprise Playbook with AI | Dreamforce 2025

Excited to see, excited to see you all here. It's both a new format for Mei and me, so it's going to be exciting for everybody here, but also for both of us. As a quick introduction, I'm a co founder at eleven Labs and we are focused on making digital interactions more human through conversational agents and for Voice. And I'm joined by fellow incredible founder Mei, CEO of rytr.

Hi everybody. We are a AI agent platform for the enterprise. This is such a cool, weird format, like disco, but no dancing. Very excited to be here with Matti, who is my fellow road warrior. We have just been talking backstage about how the last few quarters has felt like AI everywhere all at once. And I feel like every time we meet we're on a stage at a conference or backstage somewhere.

That's right from one corner of the world to another. But today in Agenda theater, of course, we'll speak a little bit more about how AI isn't just another tool. It's becoming a core part of infrastructure for the enterprise work. We'll explore this a little bit deeper. How AI is that infrastructure? What does it mean for human and AI to collaborate together? How is the future of interfaces look like and how do we do that in a transparent and a trustworthy way as the agents get deployed in production? So to kick off first question and a big question, what does it actually mean to treat AI as infrastructure instead of a tool? And how does the buyer behavior change now that you can buy that you can build the AI solutions across?

Yeah, there's a lot in that question. I think thinking about this systematically, we break down all of AI transformation as two big frameworks for our customers. One is technology and one is mindset. And on the technology side of things, what I like to really think about with our executives at enterprise companies about is the essential thing we're doing. Like if you back up to to 20,000ft, what are you doing when you lead an AI transformation or try to build an AI strategy? You're trying to take all of the intelligence that exists in your company and you're trying to encode that in AI. And that's a really important thing to do now because we very soon are going to go in an era where you are going to get agentic AI that creates intelligence ambiently, agents that interact with your employees that threaten, throw off new learnings and new intelligence. And if you haven't created a system for capturing that, it is going to be very hard to compete against people who do. So that's the one framing right and it's very important to think about the technology rollout in this way, this comprehensive end to end way, because that leads into the mindset of what we need to have leaders embodied in order to be successful. And the new leaders who are really going to be thinking about AI wall to wall, who can think about AI as leverage, are the ones who are the most creative, the ones who think beyond the silos of their teams, the calcification of their systems. And they are business leaders. I love our CIOs. We work with a lot of CIOs, but this is not a revolution that is outsourced to it. This is not a software upgrade. This is the rewiring of your entire company. And it requires business leaders to really, really take the reins. And we were on Bloomberg this morning together and I feel like our comms team was like, umay, no, don't go there. And I think I said something along the lines of like, sometimes I meet a CEO or an exec and I just want to shake them, like this is about you. They this is not about it. And those two things together, the way to approach this systematically, the mindset upgrade that's required. When we say AI is infrastructure, now that's what we're talking about.

Cannot agree more. It's very clear that especially as we think about voice about LLMs, it's not only about the research component in the deployment of AI, it's all about how do you actually link it to the existing systems for the integrations. How do you bring the salesforce, the servicenow all into the enterprise setting, connecting that brain across the organization in one clear way, then the second piece, like how do you actually build the functions to let the agents execute in their both right safeguards, ride rails, but also connecting to what you wanted to execute, whether it's a refund, whether it's an appointment scheduling, how do you actually bring that in a orchestrated way? And then I think the core other element, and that's how we think about the infrastructure at eleven Labs, is that you can create all that agenda experience, but then the main and the biggest important factor is how do you actually build and bring the agents to production. It's frequently easy to do the demo, to do a test, but how do you actually turn from that in a version way, in a testable way to then bring it to production and monitor and evaluate that over time so you can improve that agentic behavior. And across all of that stack, I think the key thing is focusing on the outcomes and really trying to understand what is that you are solving for in that enterprise setting. What is the company looking and going after? So maybe shifting that a little bit. We've seen increasing shift with our customers and with your customers from moving from pilots to that enterprise deployment. Can you unpack this a little bit more and what you said?

Yeah, I mean ROI is what everybody wants to. Who wants to talk about it? Because folks have not seen any. And I think whether it's a rollout of Copilot or a rollout of chat GPT enterprise, the vast majority of the enterprise has not seen ROI from this. And it's interesting we kicked off on infrastructure and didn't mention like you know, a, maybe not the elephant in the room but like the real reality. We have both built businesses that have bucked kind of the prevailing wisdom of you can't as a startup build a full stack solution right from the models and we're interoperable but we do build our own models, we have our own research roadmaps, both 11 Labs and Rider all the way up to the application layer. We're building a really coherent experience. And I think a big reason why our customers get ROI that far exceeds what they're getting from the rest of the generative AI basket of investments has to do with that full stack investment. And the, the ROI comes from quality. Right? I think like a huge part of what people don't talk about is things aren't scaling because they're not good enough to go wall to wall. They're not good enough to move a human in the loop or to remove multiple steps. And you need the context engineering that we're able to provide through a full stack platform. The, the quality, the context unlocks the quality and then the governance unlocks the scale. And when you build with a blank sheet of paper, a stack that's not built on the bones of something else Microsoft, then you are able to really get to that centralized observability that's required for our customers to feel like they can scale something. But you know, I'm not just going to be like, yeah, everyone scales everything. Like you can build incredible, incredible V ones that still don't go to full scale because of human reasons because not everybody is bought in or there is a lot of change management to do or the business doesn't have a tight alignment with it. So even when you get the technology right in the enterprise, there's still so much complexity to get to true scale.

I think this has been a very clear point over the last 18 months where the technology and the Quality finally moved from being able to test into bringing that into production. We've seen this across three layers. The first one, it stopped being just an innovation line in a portfolio in the business and moving increasingly into a core business line, whether it's in customer support, whether it's in sales, whether it's in marketing teams and now being core part of the strategy to the future. Second one, it's exactly like you said, it's about the roi. So actually seeing the clear results and tying this back. And then the third piece is that you can actually bring in production in a little bit quicker way and see that in your own setting and maybe to make it into a set of specific examples and the customers that we worked with. A good example is Cisco where they deployed a AI agent across WebEx which now is able to support across the entirety of their customer support their work in an incredible way. Or more recently the largest E commerce in India, Misho, where they are handling over 100,000 daily customer support calls through that work as well. And it feels extremely proud, especially here at Dreamforce, that we are also powering the voice behind AgentForce. So we're part of, of the voice for AgentForce.

That's awesome. Congratulations.

There's like a clear move towards using and working in that enterprise setting across a number of those customers. Maybe one more question on that. We've seen that shift with cloud, with mobile, but also now with AI. What's different?

What's different about AI than cloud? I think it's a full company endeavor. You know, what's possible is just so different. And you know, this cuts across the entire conversation from what do you build and what's the return on investment to how you train people, to our jobs at risk, etc. Like, you know, when you hear the analogy, oh, you know, when the calculator came out, accountants didn't become obsolete, they just did their jobs different. When Excel came out, etc. Those were tools, right? This is so much more capable of end to end creation. And I think what's different is the magic that that unlocks. Yes, there's a ton of fear, like we cannot understate that. But when you really understand what technology we're bringing into our customers, I mean we have never been able to have such a short distance between ideation and execution. Like we have CPG customers right now innovating with Rytr, coming up with new bundling, new customizations, new variations for products for Christmas, like three months away, like that never was possible. We have customers who can go from like social listening agents helping them understand what their actual users are saying on Reddit, on the web, to ideating and pro prototyping actual services that they can take live in a matter of months, not a matter of years. You've got people who are salespeople feeling like they can be marketers. You've got marketers who are shaping the future of the features that a product company builds. And you've got such incredible cross functional collaboration inside of enterprises because of AI. And none of that kind of transformation existed in cloud or in, in mobile. It felt like something one part of the company did versus a full rewrite of how we do business, how we collaborate.

The other interesting two patterns that we've seen or the shift is of course, one is with AI, you suddenly shift from using that as part of the infrastructure, whether it's mobile, whether it's cloud, but it acts on your behalf as well in your organization. So it extends a lot of your work beyond just what you input. It actually can act on your behalf. And the second thing, and I think we spoke a little bit about this earlier, to actually make it valuable, it gathers all the context around your organization into that AI setting, which is slightly different to those previous informational shifts that.

You just reminded me. Yesterday I had like five minutes between meetings. It was like 6pm I asked Ryder what GONG calls, how, what calls were recorded in GONG that day. Give me a table. The person who had the call, who the customer was with, what use cases, what positive things they had to say about Ryder in the call. And then what were some objections. And I saw one of the fields, one of the rows was with Ferragamo, a new customer of Ryder and the woman who had taken that call, a woman named Layla, who's new on the team. And I'm like, oh my God, look at this incredible call that Layla had. And I asked Ryder to send her an email from me and waseem highlighting what she did really well on the call and just congratulating her for her new call. And it was so fun to use RYTR to create that kind of connection and empathy and just like showcasing of someone's capabilities that like I could have never been a CEO that was that intimate with my team before, before our product.

That's great. What are some of the other unexpected collaborations you've seen with your, with your customers?

Yeah, you know, I have loved seeing. So we, we say we're a platform for AI native operations and our sweet spot is sales and marketing. And it's been incredible to watch a Field team feel empowered to create the kinds of experiences for their end customer. Right. Using what marketing has done for them. And Vanguard is just one of my favorite showcases of this and that marketing team built for. So your, your sales field, your field sales at Vanguard is the actual financial advisors. And you've got this marketing team that works so hard to put together journeys, right. Mattie, you've got a wonderful girlfriend, you're going to be getting married soon. Like here's all the different financial planning for you May, you've got two kids, they're going to go to college one day. Here's a path for you. And enabling those advisors to use AI to create like highly personalized advice and material and collateral using everything that marketing wants them to do, but personalizing it in only the way they can, that just was not possible. And that type of go to market collaboration we're doing across financial services, we're doing across help healthcare, we're doing across pharma. It's so awesome to see people's just brains light up like oh, I don't have to just do this little tiny thing. I can think much bigger about how to provide a better experience to my customers, to my users.

Yeah. There's two interesting shifts we've seen happen across our side. One is the user behavior shift, especially when you combine conversational agents with voice.

I'll give us examples.

Say a good example, good example is that you can shift from this being a support function. I have a problem, I'm trying to reach the customer support into this being a front facing function where it becomes part of the product experience itself. So a good work is recently we partner with Immobiliare, a biggest real estate marketplace in Italy. And their usual pattern would be I have a problem, I'm reaching out to the buyer or to the seller to get and ask the questions. And it was so successful that now it's a core part of actual embedding in the marketplace. So the first piece and the first interaction you have is actually instead of buyer, seller, it's with an AI agent. And what has actually resulted is two things. One, if you are calling about the listing through voice, you are much more likely to share a lot more information. You know, you have all the patients in the world, you can give a lot more details about your use case and then that's collated as a more holistic packet to the seller to actually give to the human seller about who you are. So the first thing was that change in the customer behavior where they are leaving a lot More information.

And that also turns a cost center into a revenue center.

Exactly.

So you're helping someone close a deal.

Exactly. And you are not anymore moving only with the problems, but you start actually enjoying the experience altogether a lot, lot more. Which has been an incredible, incredible shift. And we're seeing that across so many other places where we think to a large extent, your agent, the voice of that agent will be extension of your brand. So it won't be only a support function, but it will be actually a thing that every company, every brand in the world will start carrying as a part of interacting with their end customers. And that shift will be incredibly, incredibly important in the next, in the next few years.

Yeah, I totally agree. I think, you know, the. And we should probably talk about trust and safety as well here. And I think brand has a lot to do with that. Especially when we're trying to use AI to extend our, to extend our ux, to really extend a much more consistent brand and brand promise to our customers, then it's got to be aligned right with our actual brand and our brand promise and controlled by people in a very fine grained way. And you know, for a long time we found it really difficult from a technology perspective to help people build the granularity of voice and style into rider. Right. Whether you're producing an email or an ad, you want it to sound like Mars, like Goldman, like Accenture, etc. And the ability now via the LLMs to do this is just so extraordinary. We used to do this as post processing. Now it's built into the LLM which is really exciting and it's, you know, watching just the transformational power of the LLM scaling has also been incredible as a technology company.

Yeah, it's very important, especially as you start interacting with AI, to disclose that AI is on the other side. You build AI agents in production. It's clear that hopefully all of us will have an AI agent that can execute tasks on our behalf.

So agents will say, hey, this is a voice agent, but I'm AI.

Exactly. Or even before the interaction starts. I think one we are, we are big fans of actually walking the talk. So building a lot of agents ourselves across 11 labs and recently we deployed an AISDR. So if you are leaving your details and you want to contact.

Does it have your voice, Maddie?

Not yet. I don't think that would be the most successful one for the interaction. But we do have incredible voices across all the different languages. And the main thing that happened is that we started disclosing very clearly. You can leave the details, we'll link you to the right team. But you can accelerate that process by speaking with NIA agents and by disclosing that people actually were a lot more keen to use that as part of the process to leave more details. Totally.

And actually it's so novel too. It's like taking Awaymo. Like it is a novel.

It's actually good experience too.

Yeah.

And Waymo also needs a voice, so that's a follow up.

They're probably here.

I'm looking at a time and I think we have. Okay, we have 10 more minutes.

Okay.

How do you think that.

I'm glad you're keeping track of.

How do you think interface will shift with the agents as you think about the end user interactions and enterprises?

Yeah, I mean one of the first, when people build custom agents on rytr, one of the first things they asked us for was a, you know, a microphone icon. Right. Like I need voice input. So, you know, I think we are absolutely going to be speaking to our agents and in three to five years it's going to be like an utter sea change and I'm so excited for it. Like my favorite like just mode of technology is dune, right where it like is high tech everything but it does not look like there is a screen or a gadget anywhere. And like I think we have very similar visions and you know, work is most exciting, most creative, most collaborative when you're looking at a human being and not a screen. And you know, when I see our ability to help people organize, orchestrate across all their tools. Right. Tools that are custom built and don't have really intuitive interfaces or tools that are essential to some kind of workflow compliance that everybody hates using. And being able to have a delightful interface using it for you that you can also activate with your voice is very exciting when you marry that to routines and AI being able to proactively actively nudge you on what your routines could be at work. You know, we're looking at a very different way that the enterprise does business in three to five years. And voice, whether that is, you know, a human voice activating or a synthetic voice speaking back to you, I think is going to be a huge part of it, 100%.

It's, it's incredible how through voice you carry so much more information.

Yes.

You carry the emotions.

How many voice notes do you leave on slides back a day?

We do huddle a lot.

You do huddle a lot.

It's just quicker and you get, especially if you, if you imagine a heated discussion, it's the sync versus Async is.

So helpful to bring that between looms and voice notes like literally hundreds a day across the company because voice carries so much more information and when you're trying to convey context and our businesses, you know, we beat the big companies when we move faster, when we dissect insight from the market more astutely, when we see white space sooner and meet it faster and like it just voice is such a huge way that we convey information in our company versus like long memos and you know, we do that too but like for the fast thing it's all voice notes and it.

Definitely opens that shift in how we think about the future of interfaces where for a lot of the companies that we work with, it opens up a completely new way of interacting with end customers too. Traditionally, using and having human voice solution is very expensive endeavor. I'm sure most of you are stuck on a call with a customer service going through this ivr decision tree of trying to get to to the solution where suddenly now with AI voices, with combination with agents that we are creating, you can suddenly have that experience elevated to something that feels human, feels good. You can get to resolution in a much quicker and incredible way. And of course voice was one of the first ways we as humans started communicating as our lead and us. Ben likes to say the whole culture has been carried through storytelling, through voice and something that will carry us to the future as well. So of course a huge part for how we think about empowering businesses to interact with their customers as well. Closing up. If we fast forward five years, what do you think will feel unrecognizable about enterprise workflows?

I think we talked about just what happens with our interactions with software. I think, you know, two other things that I don't even think are going to be three to five years away. Our ability to create new knowledge via agents and the model's ability to really generalize from that experience are going to mean LLMs become your enterprise brain, LLMs become your enterprise context. And when that happens, LLMs are your system of record. And we are going to move from a human centric set of systems of record to an agentic centric systems of record. And it's going to mean a very different way of operating. The human architecture of our organizations is going to change because we have for hundreds of years built human hierarchies to overcome the high cost of getting manual knowledge work done. And we're going to get much flatter and much more orchestrated as we start managing the output of these agentic systems of record and as they interact with your goals and with your experts and with your customers. The way that people rise in an organization as leaders is going to change. It's not how many people report to you, it's not how big your budget is. Right. It's what kind of impact are you having. And I think, I think in three to five years like hundred year old companies are going to look and operate a lot more like startups because what did I just describe? I described a startup, right. So that's exciting. I hope that's exciting. And you know a lot of the customers we work with, you know the folks who lean in, who are the writer, admin, some of them are here, you know, they're the most innovative people in the company. And how do you take that energy and that way of being to everyone who works there is what I'm excited to spend the next three to five years working on.

Cannot agree more. We've seen recently. One of our proud recent engagements was working with Ukraine of government where they are trying to create a first agenda government in the world which describes a lot of the patterns that you mentioned where they are trying to bring a lot of the knowledge that would be traditionally just available to the staff in the civil service into the wider world and whether that supports or reaching out for services, from understanding the government services and being able to access that in an easier way through agents, to education, where everybody could have a personalized tutor in the future, to actually spreading the message globally around any voice in any language has been incredible. And definitely that shift towards more of a startup behavior moving a lot quicker where you try to shift a lot of the work into codifying what's the actual knowledge that people should have and then expose this through agencic workflows to the wider world.

Yeah, such a democratization. What mati do you think is going enterprise work is going to look like in three to five years?

I think it's going to be omnichannel. So it's going to combine and exactly as you think about our work shift from a combination of ways of interacting between the enterprise and the customer. So if they want to reach through text, through voice, through human, all of those options should be available and we'll see that kind of a better blending of which channel is the best one for a use case. Today we see it usually that one channel is available per one use case and that will be a big shift. Second, it will be omnimodel, whether that's visual, whether that's voice, whether that's text, will all will be blended in one place as well. And I think the core element, it's a huge part of deriving value from any of the agentic work, is actually spending the work on the foundation. So being able to link a lot of the data across the organization in a consistent manner. I think in the next few years we'll see that evolution where every organization that we work with today and the organization in the future will spend increasing time in creating a foundation that's a lot more solid, that you can derive value on. And then two, what this will unlock is that all the manual tasks will be automated and where the human input will be most valuable is the domain expertise.

I thought you were going to say you were going to be on the beach.

I think there will be an increasing amount of time to spend on the beach as well. But I think that will be that beautiful blend where you will have a lot of the repetitive work automated and easier, and then the value of human creativity, human expertise in a given domain will shine through across the enterprise work. And now we have last minute, which gives us a chance for one surprise question we each have. So maybe I'll start.

We promised each other before the session, we're going to ask each other a surprise question that wasn't in the, in the cards.

What in three years, what success look like for you and writer?

Oh, in three years. In three years we have helped the Fortune 500, like all of them, really rethink bottoms up, how they go to market, how they serve customers, how they create new products and services. And what we're talking about in terms of work being more creative, more joyful, you know, that's happening at the world's largest companies, not just the world's smallest companies. What about you?

Same question. Okay. It's very clear that how we interact with technology is shifting. AI is the biggest foundational shift for the world, probably bigger than the Internet. And I think for us and the rest of the company, we see that as an opportunity to be voice of that change, to be the voice of the AI. And we would love to, to make that possible and make it happen and be the voice of technology.

Okay, so here's my secret question for you. What voice out there would you love to license that you have not licensed yet? Dream Voice Licensee.

So I study mathematics and I love a lot of the people across that space. And we recently got to work with the estate of Richard Feynman where you can interact with Richard Feynman, learn a lot but my idea next would be Albert Einstein to have a good spectrum of characters.

Well, I bet Benioff has got some connections there. He does have the trademark, so thank you so much. Thanks, guys.

Thanks, everyone.