For AI startups, the policy landscape is expanding faster than most small teams can reasonably track. This creates a practical challenge for Little Tech: even when a startup wants to engage constructively, it may not have the resources to follow every debate in every jurisdiction.
Ben Supple, head of global public policy at ElevenLabs, joins Matt Perault to talk about his experience running a public policy function at a company that is scaling rapidly. ElevenLabs is a leader in voice AI, building products for creators, enterprises, and governments, while its public policy function is still small enough to count on one hand. The conversation offers a look at how a fast-growing AI company prioritizes policy work, builds relationships with governments, and makes the case for clear, consistent rules that startups can implement.
Ben highlights the policy environment shaping voice AI, from the challenge of tracking thousands of AI bills across state, federal, and international jurisdictions to the reason clear rules matter more for startups than for large companies with full compliance organizations behind them.
They also discuss how voice AI can improve citizen services and outcomes: replacing rigid, menu-based phone trees with more intelligent conversational agents that can resolve issues, switch languages live, and offer greater accessibility. Ben shares early proof points, including ElevenLabs’ deployment with the National Benefits Authority in the Czech Republic, where 85% of calls are now resolved by a voice agent.
For policymakers thinking about AI in government, and for Little Tech founders figuring out when and how to engage on policy, the conversation offers a useful look at what it takes to ship and scale AI products while navigating the rules that will shape them.
Topics covered:
00:00: Intro
01:00: What is ElevenLabs?
04:17: Voice AI use cases, from dubbing to customer service
06:17: The competitive landscape for voice AI
10:09: Building a policy function at ElevenLabs
11:18: Prioritizing policy work with a small team
14:31: Engaging policymakers across jurisdictions
17:29: Growing and shipping at startup speed
19:37: Human oversight and AI agents
22:12: Key policy issues for voice AI
25:00: Scaling into new regulatory obligations
29:35: State AI rules and the need for clear goalposts
31:43: ElevenLabs’ expansion in New York
33:28: Fixing the front door to government
35:40: Government use cases for voice AI
38:13: The social value of voice AI, One Million Voices, and accessibility
This transcript has been edited lightly for readability. Note, ElevenLabs’ policy team has grown since this episode was recorded. It is still small enough to count on one hand.
Ben Supple (00:00)
Most businesses, most governments have automated systems today. It just so happens that that automation doesn’t work very well. Our tagline when we announced the program was fixing the front door to government.
The technology is there today to make, you know, a much better citizen experience. I certainly value the state and local perspective, but from a very pragmatic and practical point of view, you know, it’s just impossible for a team of one, a small startup, to keep up with all the proposals.
What the EU does very well is that they have a very comprehensive and consistent framework, and maybe it’s more burdensome than other frameworks and there are challenges and costs to comply, but at least you have that goalpost and you know what you need to do.
Matt Perault (01:00)
Ben Supple, head of global policy at ElevenLabs, welcome to the AI Policy Brief.
Ben Supple (01:05)
Awesome. Great to be here, Matt. Great to see you again. Look forward to diving in.
Matt Perault (01:09)
So what is ElevenLabs? Can you talk a little bit about the company and what you guys are building?
Ben Supple (01:14)
Yeah. So ElevenLabs is, amazingly, a four-year-old company founded in 2022, launched its first products in 2023. The origin story is that you had two friends from Poland who grew up together, who studied maths and sciences, were educated in the UK, and one went to Google to work on AI research, the other was at Palantir, and they would get together on the weekends and do side projects.
And the early motivation for ElevenLabs was that when you try to watch a foreign film in Poland, they don’t have sophisticated dubbing, so typically they have a single voice actor reading every part, so like a male voice reading every part in the movie. And you ask anyone in Poland, it’s an awful experience.
Yeah, they tried to create a few synthetic, natural human-sounding voices, did an excellent job at that, and ElevenLabs was born. And I think they’ve always been focused on how do you make human technology interaction more natural, more human-centered, and voice and audio was kind of the first problem area to focus on.
So the company has, I guess, four main buckets of work. First is the foundational research. We build our own STT and TTS models, I think best in the world and, you know, you don’t have to trust me, you can look at the leaderboards. And then we have a creative platform, where you can access those models, you can access our voice library with over 10,000 voices.
You have a voice cloning feature if you want to clone your voice and then narrate an audiobook or create and scale content. A professional dubbing and production studio, so if you’re like a YouTuber in Ohio and you want to reach a global audience, you can dub your content very quickly across 30-plus languages.
And then we’ve recently launched music creation, sound effects. We have an image generation, video generation platform where you can stitch together our voices with videos. So that’s one, the creative platform. Two is the API business, so third-party integrations with our API, big companies who want to leverage our voices.
And then third, which we’re really excited about, is Eleven Agents. We found that a lot of peer companies who were building agents were using our voices and trying to stitch them together with their orchestration layer. Sometimes that led to a very clunky experience, so we thought we could do that better in-house.
So that’s a company in a nutshell. We’re about 400 people today, though that number changes by the week. And I think we were just over 100 this time last year, so we’ve almost 4X’d in a year to give you a sense of the growth. We just hit 330 million ARR, which is super exciting, and raised our Series D at an 11 billion valuation. You’re seeing a trend with the number 11, very important to the company and the founders.
Matt Perault (04:10)
Can you talk a little bit about—you gave some examples of why someone would come to you looking to use your services. Can you give a few more? What are the things that you’re seeing in the ecosystem that are driving people to use ElevenLabs technology?
Ben Supple (04:24)
Yeah. So I think voice is kind of the natural mode of communication. We’ve adapted to technology with keyboards and texting, and now I think voice and voice AI has gotten to a point where you can just have that natural conversation and unlock experiences.
Matt Perault (04:39)
What are the kinds of businesses that are looking to use some sort of voice integration and therefore coming to you and wanting to use your technology for that?
Ben Supple (04:46)
It runs the gamut. Some of the big tech players, Meta, you know, we both had our time there. For Instagram Reels, for instance, we have a partnership in India where we are dubbing all Instagram Reels in India across all Indic languages, or I think a dozen Indic languages.
Your traditional customer service use cases, we’re all familiar with calling our bank or calling our insurance company, and you’re met with this rote, very antiquated, automated experience, rigid menu base, select one for this, two for that.
So now with our voice agents, you can have that natural conversation trained on a knowledge base powered by the knowledge model of your choice. And the agent is not just deflecting you and redirecting you to a human operator at a certain point, but it’s actually resolving your issue.
So I think from a customer service perspective, game-changing. Then just like the core models, Eleven Reader, for instance, is something I use. So when I drive to work, I’m about an hour from the office, I’ll throw a policy document or a work document into Eleven Reader, and then I can select the voice I want to narrate that back to me.
So some days that’s like Michael Caine when I want a soothing British—
Matt Perault (06:00)
Michael Caine reading a policy memo? It’s a dream come true.
Ben Supple (06:03)
Matthew McConaughey if you want to mix it up a little bit. So we have a voice library of over 10,000 voices where you can kind of narrate that.
Matt Perault (06:15)
So what does the competitive landscape look like? The way that you described the value proposition sounds compelling and seems like a potentially significant market, which I assume is what’s driving your revenue growth and your valuation. What does the competitive landscape look like? It seems like something where other companies would be trying to offer comparable services.
Ben Supple (06:33)
Yeah. So super competitive, super dynamic across every aspect of our business. The model layer, a lot of the big frontier AI labs are also building their own voice models, have those in production today. From an agent’s perspective, obviously you have countless agentic products in the market.
I think we both partner with a lot of the frontier AI labs, other tech companies who use our voices kind of understand our deep focus and kind of market lead and don’t feel the need to replicate that in-house in all instances. So kind of a frenemy dynamic with some of the big tech companies and labs.
But yeah, we have to move fast. We have to ship product at a rapid clip. Certainly can’t rest on our laurels, but I think where we’ve found success is that we’re deeply focused on just a couple problem areas. We’re not trying to achieve AGI. We’re focused on AI transformation. How do you turn content into voice, into audio, and how do you make that experience as good and rich as possible?
Matt Perault (07:29)
What does that competitive dynamic feel like on a daily basis? You’re talking about larger AI native providers, so that’s probably like they’ve got legal teams of hundreds, maybe tens of people, maybe somewhere under 100 for their policy teams. Also big tech companies who have legal teams that are more like in the thousands, even potentially policy teams in the thousands. You know you’re up against that kind of competition. What does it feel like from your perspective, being at a company that may be rapidly growing, but 400 is still significantly smaller than many of the companies that you’re describing as competitors?
Ben Supple (08:07)
Yeah, from a policy perspective, there’s a lot of common cause with the bigger labs and what we’re trying to achieve from a policy perspective. I think watching what OpenAI and Google are doing from a policy and regulator engagement standpoint, what they’re doing from a policy campaign or social impact perspective is really instructive for us, and we can learn a lot from those counterparts.
I would say though, that when we’re small and scrappy, you have the advantage. You don’t have hundreds, if not thousands of policy people and 20 layers of approval before you can do something. Like when I have an idea, and one of my early passion projects was launching ElevenLabs for government. You get like five people in a room and you’re like, I think we have killer products to solve these challenges that governments face.
Like let’s get together and put something out in the market. And, you know, I had obviously worked at WhatsApp and advised OpenAI for a while and launching a program of that scale and complexity in a big company, that can take six months, if not a year, with working through the bureaucracy and all the approvals.
And so I think in some ways it allows us to move faster. The other advantage is that the Metas of the world, they have their reputation. Those views are pretty entrenched among policymakers. I go to DC and it’s like 50/50 in terms of the people who’ve ever heard of ElevenLabs. And so that first meeting can kind of set the tone of a future relationship and you can set the policy narrative, set the relationship on your terms.
And I think we’re in a place right now where we can do that on a positive basis and it’s not in response to crisis or a policy fire. So yeah, I think I haven’t seen intense competition from a policy perspective, but certainly as you know, over time, especially if you’re looking at a global remit, you can only go so deep and do so much in India and Europe and Brazil, parachuting in from California.
And I think the governments over time want to see that local commitment and have that local person on speed dial. So hopefully we’ll get to that point in the future.
Matt Perault (10:09)
You lead the team and you were the first hire of the team. You are the team.
I know obviously these discussions were before your time at the company, but do you have a sense of why ElevenLabs decided to start a policy function now?
Ben Supple (10:22)
So yeah, I report up to the legal team. Our general counsel was kind of dual-hatted doing policy and legal work, doing a bit of comms on the side. There’s a ton of excitement among governments to talk to us, to work with us on government adoption, on AI transformation, socially beneficial use. So I think he got to a point where he’s like, “I can’t be both a GC and a head of policy, and so I need someone who can come in and handle that scope.”
And I think they also have grown from a small scrappy startup to now an $11 billion company. Like we’re bigger than Snapchat. We’re bigger than American Airlines, right? So you need to build functions and build that muscle while kind of maintaining that startup culture or flat structure or limited bureaucracy.
So I think that was the early vision for policy and also to bring in people who’ve done this for bigger companies and can bring that big tech skill set.
Matt Perault (11:18)
So I think it can be a really hard stage in a company’s growth from a policy perspective when you have this outsized perception or outsized sense of the potential for the company to grow, but you’re still a small legal team, you’re still a policy team of one, and that gap can be really hard.
Policymakers are expecting you to be available to talk to them, and they’re expecting you to have lots of programmatic options available to them and points of contact from a customer service perspective, and that you’ve thought deeply about every policy position that might implicate your company. Obviously, it’s really hard for you to do that given the resources that you have available to you. So how do you prioritize? How do you decide on a Monday morning, here are the set of things that I’m going to focus on, but also the public policy function at ElevenLabs is going to get done this week?
Ben Supple (12:08)
Certainly there are days that are daunting. It’s a big scope. There’s a lot of demand. We both had the experience at Meta when you can kind of call the privacy policy expert or call the India policy team, and you had all these regional and subject matter experts at your disposal.
So I think for me it’s, how do you ruthlessly prioritize? I have four focus areas: policy development through the lens of what does go-to-market want to do, and where does the business want to grow, and how is policy an impediment to that? So if we want to do more in India, how do we build the right ecosystem?
If we want to do more in ed tech, how do we make sure we comply with child safety rules and so forth? So that’s one, kind of just showing up in our key markets. I can’t be in every country where we do business, but what I said in my initial plan was, “Let’s go to the top five. Let’s spend time in Washington and London and Delhi and Brazil and build those foundational relationships.”
Third, I is on the government adoption side and working hand in glove with go-to-market, and then fourth is social impact and socially beneficial use. Of those four buckets, I think I’ve been more focused on proactive partnerships, working with the government to help them achieve their goals, the social impact side of the house, and less on policy shaping or lobbying-like activities.
I think the other component is we want to build the right ecosystem of partners. So I bug you all the time, like, “Hey, I’m going to Washington. We don’t have access to this person. Can you help?” So we have great partnerships with you all, our other investors. We’re members of some of the big trade associations. We have good consultants in key markets. And so how do you build that early warning system, so to speak? I can’t possibly keep up with all the state and local policy proposals, but I think we have the right extended team in place that if something is breaking that impacts us negatively or positively, we’ll get a phone call and you’ll let us know where we need to engage.
So yeah, it’s been fun. It’s been dynamic. Hopefully we can prove the concept of what a policy team does and get some reinforcements over time. But as a team of one, I think we’ve done some good things and I’ve learned a ton embedding in more of a startup.
Matt Perault (14:31)
You’re based in California, and that means there are lots—and you’re a team of one, that means there are lots of policy jurisdictions that you’re not present in. You’re not physically present in DC full-time. You were mentioning Delhi. We saw each other at the AI Policy Summit in Delhi, and you were there, but you obviously are not there full-time.
Same goes for Brussels. I think that can be really hard from a policy perspective because all politics is local. Policymakers really like that there’s someone local in their jurisdiction who they can call when they’ve got an issue. And I think it can be particularly challenging for companies that do have a fairly significant amount of public visibility, but still have a pretty limited local footprint.
I remember at Meta when we were kind of in the news every day but didn’t necessarily have a policy manager in every single country or even a small fraction of countries, and that delta could be really frustrating to policymakers. It seems like it probably is really exacerbated for you.
Again, you’re trying to have as sizable a presence as you possibly can, but you’re not going to have the kind of presence that larger companies are able to have. Does that affect the nature of the back and forth that you have with policymakers? When you show up in Delhi and you’re talking to them, how are they thinking about your company, and how do you deal with the local presence issues?
Ben Supple (15:45)
Certainly working across time zones, I think it’s a 13 and a half hour time zone difference between California and India and asking policymakers to talk to you at 7:00 p.m., 8:00 p.m. at night is not ideal. At the same time, they see that we’re a three-, four-year-old company and growing fast and there are a lot of companies I’m sure in your network that are much older, have been around for longer, and don’t yet have a policy function.
So I think in some ways it’s very forward-leaning at the stage of the company. We’ve kind of built that muscle. The other thing I would say is that we’re still a small team. We’re not super stovepiped yet. So if there’s a person in Poland who’s interested in policy, but maybe he works in finance or works in marketing, in some cases we’ll just tap someone else in who’s not a policy person per se, but interested and can engage on the company’s behalf.
And then as I mentioned, how do we get the right consultants, work with the right trade associations? How do we make sure that those trade positions are representative of our view and not just the big tech view? But yes, definitely a challenge to work across all those jurisdictions, especially in this current geopolitical climate where there’s such an obsession with sovereign AI, and you have to kind of claim your allegiance to one country or region.
So yeah, showing up and showing commitment and that each country and region is special while not actually having a physical presence is certainly a delicate balance.
Matt Perault (17:29)
Ben, you’ve talked a bunch about growth. You’re at $330 million in reported revenue in 2025. You got there after five months. Five months before that, you were at 200 million. This is an incredibly rapid pace of growth. What’s it like to be at a company that’s growing that fast?
Ben Supple (17:48)
It’s a lot of fun. Certainly Meta, we grew fast, we shipped products fast. I had some exposure to OpenAI, and I’d say this is like a step change, more intense and faster. And some of that is just because we’re small, we’re a startup, we don’t have layers of bureaucracy, as I’ve mentioned.
And I think the competitive environment, right? Every week people are shipping better models. I think people inherently understand that to win in AI, you need to be multimodal, you need to have a solid voice layer. So yeah, I think it’s intense in terms of the competitive dynamics, in terms of the team size.
But we also have a killer product, and I think the product kind of sells itself. I had this experience at WhatsApp where I think WhatsApp achieved like a billion or a billion and a half users before they hired their first marketing person. And I’m like, “That’s kind of cool.” It sold itself to over a billion users before you needed a marketing team.
And I feel like ElevenLabs is in that same vein, where it’s like once you’ve experienced it, once you’ve had a conversation with a voice agent who understands everything you’re saying when you’re driving to work with a lot of background noise, and they not only understand you, they’re emotionally aware, and they understand when you get frustrated, and they change the cadence at which they speak.
And they can actually resolve the issue for you without having to redirect to someone else and wait for 30 minutes. It’s game-changing. It’s kind of like the ChatGPT moment. It’s like I didn’t think I needed this, but now that I know it exists, I can’t work without it. And I feel like our technology is in that category where once you’ve had some of these conversations with ElevenLabs voice agents, it’s game-changing.
And you’re like, “You know what? I don’t know if I want to be redirected to a human operator.” This was really cool, and it totally understood what I needed.
Matt Perault (19:37)
It is so funny what you say about redirected from an AI tool to a human, because there’s so much conversation now in policy debates about the importance of humans in the loop. I think we’ve seen this over a long period of time, but increasingly, at least at the state level, a lot of the bills that I’m seeing are mandated human to be involved in decision-making.
And that might be good in some circumstances, but I think there are any number of different circumstances where maybe a human in the loop is a suboptimal experience. How do you guys think about that? It sounds like part of the idea of the product is to take some number of conversations that would be between humans and move at least some percentage of them some of the time to between a human and an AI agent.
Ben Supple (20:20)
I think most businesses, most governments have automated systems today, so the front door to government or the front door to customer service at a business is already automated. It just so happens that that automation doesn’t work very well and it only deflects and redirects, and it puts a big strain on human operators.
I think there’s always last mile support where you will need a human operator to jump in and to work with you hand in glove. But what we found, and we’ve all experienced this, is in peak times when you’re calling your airline, you’re calling your tax authority, there just aren’t enough butts in seats, so to speak, enough human operators to address the demand.
And so imagine if you had an AI assistant that could resolve the basic issues that 50% of people are calling about. But yes, I definitely think you want that human oversight in the system design. You want that kind of phone-a-friend option to a human operator. I think that’s required, both from a policy perspective, but also a lot of companies, like when I was at WhatsApp, you had to have a human escalation path built into the product flow, otherwise you couldn’t launch it.
So we certainly value that, but I think what we’ve seen is that, you think through government use cases, for instance, and how many people don’t apply for benefits who are eligible because they get confused, they get frustrated, they don’t want to wait for two hours. And so imagine if you had an AI assistant who could work with you and resolve 50% of those issues and get people the outcomes and the services they need.
Matt Perault (22:12)
So you lead the policy apparatus, so let’s dive in a little bit deeper on policy issues that you’re facing. What are the policy topics and dynamics that are keeping you up at night right now?
Ben Supple (22:22)
The first thing I would say is I think when people hear “AI startup”, their mental model goes to OpenAI and Google and Meta and the litany of policy challenges we’ve all read about in the headlines. So first, I think it’s important to kind of situate us in the AI stack. We’re not, as I mentioned, racing toward AGI, right?
We’re focused on transformation scenarios, transforming speech to text and text to speech. So think of us less as an OpenAI where you’re worried about catastrophic risk. Can you use this tool to create a bioweapon or change national security outcomes? And more like an Adobe for audio, right? Where we’re providing that creative suite.
But the spectrum of policy risks is very unique and very different, and I think in some ways more limited than some of the big frontier labs. So what are we focused on? I think issues on digital replicas, voice cloning, things like the No Fakes Act at the national level, Elvis Act at the state level.
Matt Perault (23:20)
Are you generally comfortable with those current frameworks, or do you think that those frameworks underappreciate either elements of a company like yours or the kind of technology you’re building? Are they appropriate for voice AI generally?
Ben Supple (23:33)
I think we need to study each example more, and those I know are still working through the process, so I think at the right time we’ll do the deep dive. Generally, I think we’re pretty confident in our safety stack, and at the end of the day, irrespective of the policy environment, we want to be an enterprise-grade company that can sell to the biggest companies in the world, including in regulated industries and government.
And to do that, you have to be high trust and above board on all the compliance frameworks. So generally, I think we’re in a pretty good position with the safety approach and the compliance approach we’ve taken. But yeah, in terms of other issues, think robocalls, AI disclosure and robocalls.
We have agent products that can actually proactively reach out to people and execute those calls. There are privacy dynamics where voice prints are considered biometric information that we pay attention to. And then, of course, just keeping up with the broader debates on a federal standard versus a patchwork of state and local laws.
And I think for us, we’re in favor of certainty and consistency, and wherever that policy debate lands, we just want to make sure we know what we need to do to comply and can continue to run fast.
Matt Perault (25:00)
So I’m curious about how these dynamics play out in terms of how you’re situated vis-a-vis some of the larger competitors that you discussed. And when you talk about revenue and revenue growth, unfortunately, the thing that I have to think about immediately, because I lived it in a fairly intimate way, is the revenue threshold in SB 53, which is California’s mandatory disclosure law.
And so from our point of view as a firm, there are a lot of concerns about SB 53, but one of the things that we thought was good about it is it uses a revenue-based threshold to try to separate between larger providers and smaller ones. And the reason we think revenue is a good way to draw that line is that at some point, when you’re taking in enough revenue, you could hire a general counsel, or you could hire a head of policy, or you could write checks to third-party auditing firms to help you with disclosures.
But you’re headed toward this revenue threshold. You’re headed toward the $500 million in annual revenue, and you still have a small legal team, you still have a policy team of one. Obviously that list that you just went through is an incredibly broad list of various different priorities, and it would take you a long time, a lot of effort to make sure that you’re touching all of them.
It sounds like at some point in the future, you’re likely to hit this heightened disclosure threshold, and that will put you in the same category as the companies that you described as potential competitors, like the OpenAIs and the Anthropics, but also the Googles and the Metas. So can you say a little bit about how that sits with you? How do you think about this as you go from a small footprint to a larger footprint that will come with heightened regulatory obligations?
Ben Supple (26:25)
I think we’re fast approaching that threshold and that’s a good thing. And the challenge with growing so fast is that you become subject to different regulatory regimes. So yeah, I think as we achieve those thresholds, we’ll have to expand the team appropriately.
I should say that I’m a team of one for public policy, but we also have a pretty big safety team. We have a compliance team. We have a relatively large legal team. So we do have a cross-functional structure that can help us comply with more burdensome regulations.
And the other thing I would say is, as a company with European founders where Europe is an exceptionally important market for us and you have very robust and burdensome, in some cases, regulations, we’ve worked hard to comply with all European regulations. And so I think a lot of these disclosure requirements that would be required in California and elsewhere, there’s already kind of a product roadmap to comply with those, sometimes not in response to that local bill or a US proposal, but what we’re doing internationally.
Matt Perault (27:40)
I think your dual presence in Europe and the United States does make you unique, and it’s at a time where there’s lots of tension between the US and Europe on trade issues and digital issues specifically. I assume that puts you in a complicated position, although maybe it also gives you opportunities on the policy side as well. Can you talk about that? How do you navigate those tensions?
Ben Supple (28:02)
It’s really interesting. We’re very much in some ways a European company with Polish founders who care deeply about Poland and EU issues. They were educated in the UK and kind of conceived of the company in the UK. That’s where we have one of our largest hubs.
We are incorporated in the U.S., and have a huge team in the U.S. So I think very authentically we are more global than a lot of our competitors in that we have those roots in Europe and roots in the UK and the US. Often the growth path is like companies will start in California, they’ll build for the US, and then over time they will localize for international markets.
For us, I think Europe has always been a business priority. The UK has always been a business priority. So I think we’ve had to build that compliance muscle in the first instance, not because we wanted to localize in the future, but this is where we are today and where the center of gravity of the company is.
So I think in many ways it works to our advantage. I saw this at Meta and other companies where they felt very US-first from a policy perspective and certainly there’s now a lot of pressure to choose sides and where do you want to build. But I think we’ve been able to thread that needle very effectively and have been able to move fast internationally without having to go deep in one country before we’re allowed to go elsewhere.
Matt Perault (29:35)
So navigating jurisdictions in a complicated way is part of U.S. and Europe policy work right now, but it’s also part of federal policy work vis-a-vis state policy work in the United States. I’ve heard different numbers, but 1,400, 1,800 bills related to AI introduced at the state level.
Again, you’re a small team, a mighty team, but a small team. How do you guys go about trying to make sure that you are not disadvantaged by policy developments at the state level?
Ben Supple (30:05)
Tall task, no doubt. I’d love to say that, yeah, I’m across all the proposals and I’ve gone deep in everything in California and Colorado and elsewhere, but sadly there’s just not enough time in the day to do so. I think our approach is really partnering, as I’ve mentioned, with the TechNets of the world, working closely with you all and other partners who have those state-level trackers and can give us a heads up when something is percolating that might impact us.
But as I said before, I’ve engaged a lot with the California government. I think they’ve taken a very thoughtful approach. There’s certainly room for local nuance and how these tools impact Californians in a way that may be different than other states. So I certainly value the state and local perspective, but from a very pragmatic and practical point of view, it’s just impossible for a team of one, a small startup, to keep up with all the proposals.
And what the EU does very well is that they have a very comprehensive and consistent framework, and maybe it’s more burdensome than other frameworks, and there are challenges and costs to comply, but at least you have that goalpost and you know what you need to do. Whereas here, there’s this patchwork and there’s that debate between the federal standard and the state and local.
And so I think we’re in favor of coming to a good compromise solution and giving us that kind of clear goalpost of what we’re trying to achieve.
Matt Perault (31:43)
You guys recently announced an expansion in New York. What was the rationale behind that? And I’m curious if that means you’ll be doing more policy work in New York specifically. It’s a state that’s important from a policy perspective. They moved a transparency bill that ended up looking a lot like California’s, actually. They moved that last session. There’s a lot of stuff on the table now. So I’m curious about the expansion and then how that’s going to affect your policy engagement as well.
Ben Supple (32:12)
Yeah, so New York has always been our primary U.S. hub. Obviously a great talent pool in New York, an incredible city, thriving tech scene. Some of it is practical in that, with a lot of folks in Europe and the UK, managing across time zones pulls a lot of people eastward.
I think fundamentally we outgrew our office in New York and had more demand and needs than we could satisfy, and were able to work with the government to secure a bigger office space and worked with them to articulate our growth plans, which includes hiring over a couple hundred folks, investing in more tech and engineering talent.
I think we committed to a sizable R&D investment and had the partnership of the governor and the state throughout that process. So yeah, we’re excited to go deeper in New York. It’s always been kind of the center of gravity in the US for us.
But I would say that from a San Francisco perspective, we’re also growing at pace in the Bay Area and have people spread throughout the country. So we hope to have more headlines like that in the future talking about our exciting growth in different jurisdictions across the country.
Matt Perault (33:20)
You also released ElevenLabs for government recently. What was the idea behind that program?
Ben Supple (33:28)
So as I mentioned, this was something near and dear to my heart. When I was at Meta, I kind of split time between public policy and then government partnerships and government sales. Our tagline when we announced the program was fixing the front door to government. And I think in most instances, when you have an important need of your government, at the local level, you’re picking up the phone and you’re calling, right?
You’re calling the school district to see if your kid’s enrolled. If you’re struggling to figure out how to get a business license, you’re calling the business department, the tax authority. If you’re trying to navigate benefits, let’s say you need food assistance or housing assistance, you’re in a moment of crisis or need, you want that human touch.
You want to call someone and have a conversation and not navigate complex web forms and have a conversation over chat with a virtual assistant. But when you enter that front door, as we’ve discussed, it’s a horrible, antiquated, in most cases, automated system, menu-based, rigid, redirecting, deflecting, not actually taking action and resolving any of your issues.
And I think the technology is there today to make a much better citizen experience, where you can call and have a virtual assistant speak to you with human touch and actually be trained on a knowledge base to take action and resolve a large percentage of issues that you need.
And that’s a benefit from a company perspective, but from a citizen experience and citizen outcome perspective, how many people give up and don’t get the benefits and the support they need from governments because they can’t navigate the process?
In California, for instance, I think we have like 200 spoken languages in the state, huge Spanish-speaking population. With our agents, you can have a conversation, live language switch. The agent can even pick up a local accent and speak to you in Mexican Spanish or Colombian Spanish.
So there are just incredible capabilities in terms of personalization to help improve those citizen outcomes and avoid the high abandonment rates that we see too often.
Matt Perault (35:40)
I know the program launched pretty recently. Are you seeing interesting use cases so far?
Ben Supple (35:47)
Yeah. So we have a few existing government partners in Ukraine. We’ve worked with the Ministry of Digital Transformation. Twofold project. You had the minister who created a digital twin so he could actually deliver updates in his authentic voice, and then they have a citizen engagement platform called the Diia app, through which you can access dozens of services that was limited to text, and so we’re powering the voice layer.
And then a few examples of call center digital transformation. So in the Czech Republic, we work with the National Benefits Authority. So that first line of defense, the automated system is our voice agent, and amazingly, I think 85% of calls are resolved by the voice agent. Used to be that 100% had to be redirected to a human, now only 15%. So super powerful.
As you mentioned, we just launched it, so we’re in a bunch of early conversations. I think there’s a lot of excitement. But in my mind, I saw this evolution when I was at WhatsApp, where you had almost no governments in, say, 2019 delivering services through chatbots. By the end of 2020, and a lot of this was propelled by the pandemic, and call centers were overwhelmed, and they couldn’t keep up, and you needed to reach people more proactively, we had half of product volume coming through from governments by like May 2020 in a five-month period.
Zero to 50% of product volume in a five-month period. And I think thousands of governments are now delivering services through text, AI-powered text chatbots. Every single one of those, I think can be improved by building a voice layer, right? And think of the options for personalization, as I mentioned.
You can only personalize so much over text. With voice, having that emotional intelligence, having local accents, switching across languages. So I’m biased, of course, but I kind of see voice agents as that next big thing for governments. So super excited to dig in. We have some early proof points, but a lot of work to do.
Matt Perault (38:13)
Seems like an awesome project. We’ve talked a little bit about the growth of the business. We’ve talked about the use of the product for government. Can you talk about just the general social value of voice as well? What is the potential of voice for improving our economy and our social life?
Ben Supple (38:28)
Yeah. So super excited about the social value and exploring socially beneficial use. Obviously from an accessibility standpoint, there are a lot of people who prefer to engage with voice and maybe they can’t navigate complex web forms and so forth. And so it gives them an extra tool in the arsenal to engage with that content.
We have some existing programs in place. So we launched a One Million Voices program, which is incredibly inspiring. The concept there is that if someone has suffered permanent voice loss due to an accident, due to a medical condition like ALS, they can use our tools to create a synthetic version of their voice, and we give them a free license in perpetuity to do that, so they can communicate naturally and authentically in their own voice.
So we launched that program relatively recently and are excited to work with governments and nonprofit partners to scale that. One quick anecdote. We had a big summit in London recently, and I think for the first time we worked with an individual who was a singer in a folk band and worked with him over several weeks, maybe months, to recreate his singing voice.
And we brought him on stage with his old band and powered an AI singing voice of his, which then got picked up by national news in the UK and created quite a splash. So great foundation, I think, on the socially beneficial use side of the house, and I think there’s so much more we can do.
And how do we partner with governments, right? So there are AI for science moonshots and AI for education and healthcare, and how do we incorporate our technology across those verticals and help governments and civil society achieve those moonshots using voice as a key lever in that process?
Matt Perault (40:15)
Inspirational. Ben, thanks so much for joining the AI Policy Brief.
Ben Supple (40:18)
Thank you. Great to be here.
This newsletter is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. Furthermore, this content is not investment advice, nor is it intended for use by any investors or prospective investors in any a16z funds. This newsletter may link to other websites or contain other information obtained from third-party sources - a16z has not independently verified nor makes any representations about the current or enduring accuracy of such information. If this content includes third-party advertisements, a16z has not reviewed such advertisements and does not endorse any advertising content or related companies contained therein. Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z; visit https://a16z.com/investment-list/ for a full list of investments. Other important information can be found at a16z.com/disclosures. You’re receiving this newsletter since you opted in earlier; if you would like to opt out of future newsletters you may unsubscribe immediately.
