The World’s Best AI Voice Company Chose Humans Over Agents

The world’s leading AI voice company, ElevenLabs, needed to scale their outbound. They can build agents that sound human and act human better than anyone on the planet; it is what they do. So when they needed to scale their outbound, like, what did they do? Did they build an army of AI agents? Nope. They hired humans. And they are crushing it.

Let that sink in. The company most equipped to automate outbound chose not to. So the backstory here is, like, about a year ago, Jonathan Shimony came in to lead the go-to-market role at ElevenLabs. And at the time, they were running about 95% inbound. And he was tasked with building an outbound engine and getting that scaled up. And he did—with humans making phone calls—and now they’re at 46% outbound source pipeline and they’re climbing. All humans.

And he’s not the only one. I’m talking to go-to-market leaders and CROs at companies with massive AI development budgets, of incredible AI talent on their actual payroll right now, and I’m hearing the exact same thing from every single one of them. What you’re seeing on your LinkedIn feed—like these "my AI agent booked 46 appointments a day," like, just basically "hit the easy button, press go, AI can replace all your SDRs and all your BDRs and all your outbound and all this shit"—it’s not the fucking reality. You’re hearing it from people who sell AI agents, but the people who are in the trenches—boots on the ground—the people who are actually responsible for getting the results, they’re telling a very different story.

It’s the same story that I’m telling, and it continues to get reinforced. And I actually just did an interview with another founder, Nick Nicolo. He’s the founder of an AI company that helps augment human outreach with the outbound engine. So like, think using AI for dialing, for data enrichment, for cleaning, for a lot of the stuff that surrounds the SDR or the BDR function. All of his stuff—the AI that they’re building—all makes human SDRs better.

So he’s building AI tools for a living; he’s come from the sales world, SAS companies, and he will tell you: the only use of AI getting real results in outbound right now is augmenting humans—amplifying the work that they are already doing, making it better, making it more effective, making it more efficient, not replacing them.

Why? Why isn’t AI the answer for outbound? And it comes down to basically three things. One is—and before I get into it, there’s an old saying about the gold rush: the people who got rich weren’t the ones who were mining for the gold; the ones who made a killing were the ones that were selling the pickaxes and the shovels. The same thing is happening right now in AI. People, they see a gold rush in AI and they want to sell you the tools, because that’s where the money is. But that doesn’t mean the tools are actually getting results.

Okay, three reasons why, like, AI is not the answer to replacing your outbound salesperson or what you’re generally hearing right now.

Number one: inbound and outbound are psychologically different, and AI is phenomenal for one—as even a first touch point—and not for the other. Inbound, somebody’s coming to you. They’ve already got intent; they’ve searched, they’ve shopped around. They think they know what the solution is, and there’s at least some small chance that they think the answer to their problem or the solution to their problem is what you sell. If you’ve got an AI agent that’s there to answer questions, that’s there to give them the information that they need to make their buying process easier, to make it take less time, to make it more efficient for them, that’s a service. That’s saving them time. That’s a great use of AI.

Outbound is the opposite, though. People don’t want to talk to you in the first place when you’re doing outbound. You know this. There’s a reason nobody likes making cold calls and going outbound. The people that you’re talking to, they’re not looking for you, they’re not looking for your information, they’re minding their own business, and you’re interrupting it. That’s why outbound is hard just in general. Now imagine someone steps out of the office, they’ve got a little time to pick up their kid from school, and then they get a call from a robot. Or they’ve got an hour between, like, a Zoom marathon session all day, and your AI agent rings them up, interrupts the hour.

How do you think that’s going to go? How do you think people are going to respond to that? It’s hard enough to get conversations when you’ve got humans calling. When you’ve got a robot—even when they have humans, they treat them like robots. What do you think they’re going to do, or how they’re going to respond when an actual robot calls them? "Like you didn’t even take the time to have a human annoy me; you built an agent to do it."

The same dysfunction that causes outbound to fail with humans is exactly what’s going to cause it to fail with AI. So think about it: if you want to build a really great AI agent for outbound, what do you need? You’ve got to have clear playbook for them to follow, you’ve got to have the processes, you need time to train it properly. It’s not going to be excellent on day one; this is AI 101. You’ve got to be patient, you’ve got to iterate, you’ve got to make changes. Then you’ve got to have regular sessions to quality-check it, to coach it, to manage it.

Does any of that sound familiar? If you’ve hired salespeople—and I’ve worked with hundreds of small businesses that are trying to hire salespeople and trying to hire SDRs—those efforts fail for the same three reasons:

They don’t have a clear playbook or process to follow and they won’t and they won’t invest the time to create one.

They don’t have a plan to train that person and they won’t invest the time to do it the right way.

And they don’t take the time to manage, to review calls or coach.

And the foundation is fundamentally the same. An AI agent isn’t a shortcut around doing the work; it’s just a more creative way to skip it and not get results.

And number three: the math doesn’t work for most businesses who want to turn on AI to scale up their outbound. AI has a very clear advantage, and that’s volume. If you have an AI agent, it will make more calls per hour, it will use more hours per day than a human ever could. Hands down, AI is a winner when it comes to volume and activity.

But when it comes to actual conversion, it’s actually less efficient. We’ve measured this at my company, MSP Sales Partners. We manage 65 SDRs who are doing outbound and we’ve done plenty of work with AI. We’ve tracked the data. I’ve also seen a ton of these AI systems through my investment company, Repeatable Revenue. So when we’re evaluating them, we get to pop the hood and take a look at what they’re working with. So we’ve seen many, many, many instances of AI.

And what we find based on the research and the data is that a good human SDR is going to convert at six to seven times the rate of an AI agent on a per-lead basis. If you are measuring this on a per-lead basis—you’ve got a thousand leads to go do a campaign with—the good human doing the outreach will convert at six to seven times higher based on the number of people out of that list.

If you’re operating in a really defined market—so say you’ve got a specific geography that you work in or you’ve got an industry, you’ve got a specific vertical—and by the way, you should have that because if you’re not, your outbound’s probably not going to work anyway—if you turn an AI agent loose on that list, you’re going to burn through your market at half the conversion rate or less. Your total addressable market is literally half as valuable as it was before you turned the agent on.

And that’s before the incoming wave of regulations, of disclosures, crackdowns—all kinds of other shit that’s... there are going to be disclosures that say, "Hey, you can’t use AI to imitate a human," or you have to disclose when this is going to be AI. All of this shit’s coming. And this is—so the conversion data I’m giving you is before all of that.

You may be thinking, "Alright, Ray, you’re just old school; you’re an anti-AI curmudgeon," right? And that’s fair. Okay, except my team has access to team-level AI accounts with Anthropic. We’ve got team-level accounts with OpenAI. I’ve got an AI agent right now cleaning and enriching my CRM. I’ve got another one that takes my podcast and automatically pulls the podcast, gets the transcript, and sends that transcript using automation out, using stuff that’s been trained on my voice to give us drafts of LinkedIn posts, of tweets, of YouTube scripts, of newsletter drafts—all using Make and AI.

I’ve used AI to coach me through some really high-level decision-making. I’ve used it in building operating systems within my business. I’ve used it to coach through personal health challenges. Listen, I am a fan of AI. But this isn’t about being for or against AI; this is about what gets you actual results. And if you listen to the people whose jobs actually depend on the outcome, on the result—like the sales leaders who live and die on whether the outbound engine or system that they’re creating or they’re running is actually driving real pipeline—they are not saying AI is my answer.

In virtually every business doing outbound right now, AI agents are a shiny object. They’re a distraction. And the leaders that are being told, "Hey, great job on outbound, now go build an AI agent to replicate it all so we can get rid of these salespeople," instead of using that time to scale what already works, it is killing results. I’m seeing it across multiple businesses. That’s why people who are actually in the trenches aren’t saying "AI to replace all of my salespeople is the next way." And you’re only hearing that from people who are selling AI solutions or benefit as a result of maybe they’re promoting big events like "Hey, come see how we’re using AI to replace all our salespeople, buy our tickets," or "buy our software," or "buy our [whatever it is]."

And like, listen, if Jonathan Shimony at ElevenLabs—the head of go-to-market at the world’s leading AI voice company—is choosing humans over agents, if that’s not enough evidence for you, well, okay... but don’t say you haven’t been warned. Hope it helps. Adios.