Eleven Labs Went From 600M+ ARR in 41 Months. When the AI Agent Closes the Deal, They Still Pay the Human. With Carles Reina, First VP of Revenue | Saa Str AI
Overview
AI VC AI Mentor: Digital Jason + Amelia AI Startup Benchmarking
AI Agent Playbook Free e Books
e Book: Hiring a Great VP of Sales
e Book: Raising Capital
e Book: The First $1m ARR
Details
University All Posts Podcasts The Top CROs VC Fundraising Top Videos Q&A Best of Saa Str #1 Bestselling Book Search Everything Join the Community
Free e Books
e Book: Hiring a Great VP of Sales
e Book: Raising Capital
e Book: The First $1m ARR
AI Annual 2026 Events Overview Sponsors
Event Sponsorship
Media Sponsorship
Digital AI Day 2026 (Free) Speaker Submissions Speaker Requirements Overview
Eleven Labs Went From 600M+ ARR in 41 Months. When the AI Agent Closes the Deal, They Still Pay the Human. With Carles Reina, First VP of Revenue
by Jason Lemkin | Artificial Intelligence (AI), Blog Posts, Podcasts, Saa Str. Ai, Sales
Carles Reina was employee #4 at Eleven Labs and its first go-to-market hire. He was also its first investor. He spent the first nine months selling enterprise deals by himself, then built the entire commercial org, and has now moved full-time to Baobab Ventures, the $15M solo GP fund he raised last year.
He came on Saa Str AI’s CRO Confidential with Sam Blond, now founder and CEO at Monaco and previously CRO at Brex and Zenefits. So this is two people who have built revenue orgs comparing notes rather than one person presenting a case study.
That is 41 months from zero. The growth rate went up as the numbers went up. The New Normal for the best in AI + B2B.
Five takeways from Carles on the road from 600M+:
When an agent closes revenue on an account, still pay the human who owns it
Give the product away free to the exact segment your competitors depend on
Give a market its own thesis, its own channel mix, and a defined 3-to-6 month result before you launch it
Set quota at 20x base salary, pay uncapped, and publicly commit to lowering it if the market says you were wrong
Pay commission only on recurring contracts, and pay $0 on a POC no matter the size
20x base salary is where he set quota.
167% average quota attainment across the entire go-to-market team, every quarter.
600% of quota for the top AEs, with a lot of the team running 300%.
10%+ of enterprise revenue traced back to a program that gave the product away free.
#1. When the Agent Closes It, Still Pay the Human
Two and a half years ago, at a company offsite in Switzerland with under 30 people, Carles pitched the Eleven Labs founders on building an AI go-to-market org: an AI SDR, an AI account executive, an AI customer success manager.
The answer was no. The technology is not there, keep hiring people, go faster.
He kept pushing. A year later he got one developer dedicated to go-to-market, and started building.
The objection from his own team was the obvious one. Am I getting replaced?
Two demonstrations and one comp decision settled it.
The demonstrations: an AI SDR that answers an inbound and calls back immediately converts better than a human replying within 30 minutes, and no rep actually wants to be the person guarding the inbox. An AI customer success manager working upsells across the SMB and mid-market longtail unlocks revenue nobody had time to chase.
The comp decision is the one most companies skip. When the agent unlocks revenue on an account, Eleven Labs still paid commission to the human who owned that account.
Carles recommends this to his portfolio companies now. You are paying twice, once for the agent and once for the person. That is the price of removing the friction. The alternative is a rep quietly working against the system that is supposedly helping them.
#2. The Grants Program: Free for 3 Months to Startups Under 25 Employees
For the first nine months after Eleven Labs shipped, Carles was the only person selling. His competitors at the time were other small model companies going after developers and startups.
He went on holiday with his wife. His stated rule for holidays: figure out how to kill the current set of competitors.
What he came back with was a grants program. Startups under 25 employees got Eleven Labs free for three months. They made a lot of noise about it publicly. They ended up giving out tens of thousands of grants in a short window.
The result: for a long stretch afterward, over 10% of enterprise revenue came out of the grants cohort. Companies that started building on Eleven Labs for free, grew, and got upsold.
Carles is direct that this was a gamble and they had no idea if it would work. What made it work was the targeting. His competitors were the same size as Eleven Labs, funded at the same level, and living on developer and startup adoption. Three months free to every startup under 25 people took that adoption off the table for them, and they could not run the same play back.
The cost was three months of usage per account. The return was over 10% of enterprise revenue.
#3. Every Market Launch Got a Written Thesis and an Expected Result in 3 to 6 Months
Eleven Labs launched enterprise across the US, Europe, Japan, India, Korea, Brazil, Mexico, Colombia and the Middle East. Each one got its own answer to three questions before launch: why this market, what channel mix, and what result do we expect in the first three to six months.
The channel answer changed by country. Core markets got direct sales. Others went reseller-first, and tax decided it. Withholding taxes and local invoicing made direct selling worse economics than handing a reseller margin. A finance constraint set the go-to-market structure, market by market.
Writing the expected result down before launch is what gives you a date to score it on. The alternative is entering a market, staffing it, and arguing eighteen months later about whether it is working.
#4. Quota at 20x Base Salary, With a Public Commitment to Lower It
The old B2B benchmark was roughly 5x. Pay a rep
Carles set 20x against base.
He says the number came from two places. First, his own nine months as the only seller, extrapolated. Second, a decision about what fair meant in a market moving this fast.
The first two reps he hired told him there was no way. What he told them:
I don’t know if you’re going to get there or not. What I can promise you is that we will make it fair. If in the next 3 to 6 months we see that you’re not getting there, I will lower the commission, I will lower the targets. If you get there in six months but in a year’s time the markets have changed, we will reevaluate as well and we will drop it.
I don’t know if you’re going to get there or not. What I can promise you is that we will make it fair. If in the next 3 to 6 months we see that you’re not getting there, I will lower the commission, I will lower the targets. If you get there in six months but in a year’s time the markets have changed, we will reevaluate as well and we will drop it.
Both of those reps signed contracts in their first week.
Three and a half years in, the outcome he describes: account executives running 600% of quota, a lot of them running 300%, and average attainment across the entire go-to-market team of 167% per quarter.
He also honored the other side of it. In quarters where a market’s fundamentals changed, they gave quota relief, in some cases 50%, across the board in the affected markets.
His preference stated plainly: a smaller team that is very well compensated over a bloated team that is missing quota, complaining, and demotivated.
20x worked at Eleven Labs because the product was pulling demand faster than humans could process it. Sam’s warning on the episode: there is no rule in AI, and copying 20x onto a business without that pull is how you get a floor full of reps at 40% of quota in November. The part that copies cleanly is the promise he made alongside the number, and the 50% quota relief he actually granted when a market turned.
Eleven Labs paid commission on recurring revenue contracts only. Sign a two-month or three-month POC and the rep got nothing. Carles is explicit that this held whether the POC was
His reasoning to the team was an equity story argument, not a compensation argument. His framing: roughly
Sam brought the same lesson from Brex, arrived at from the other direction. Brex made money on card spend, so it was usage-based years before that was normal. They started by incentivizing activations and meetings. Revenue ripped when they moved reps, and eventually SDRs, onto revenue itself.
Both of them landed on the same test. Carles reported four ARR milestones on this episode and no meeting counts. Whatever number you put in the board deck is the number the comp plan should pay on.
#6. The Two Things He Would Have Done Differently
Asked what would have gotten them past $1B by now, Carles named two, and both are hiring decisions.
Sales enablement and GTM operations, much earlier. He describes it as the item that gets overlooked, hired long after the team is already big, at which point you are retrofitting onboarding onto people who already made up their own methods.
Sam’s answer to the same question was identical, from Zenefits. His line: he has never heard a sales leader say they went too big on revenue operations. His explanation for why it keeps happening is that sales leaders come up through selling and are execution oriented, so ops is the skill set they are least likely to hire for themselves.
More senior reps, sooner. The default startup instinct is young, hungry, driven, cheaper. Carles’s counter is that the default means starting from scratch on every account. A rep with 20 years in the market walks in knowing the procurement people. He wanted a mix, and he wanted the senior end of it earlier.
#7. What Agents Are Still Bad At, From Someone Running Them at Scale
Carles’s read on agents in sales, from having built them: they are not good yet at anything relationship-driven, and they are getting better.
His explanation of why: LLMs work off the distribution and return results inside the distribution. Selling well means operating outside it. Everything in a sales motion that is genuinely differentiating is by definition not the median response.
What they are excellent at today: analyzing large amounts of data, researching accounts, building the TAM, scoring accounts against criteria, overlaying signals, drafting the message off the signal, updating records, working 24/7.
Sam’s sorting rule: if you are spending your time finding companies to add to the database, researching people for a reason to reach out, dropping people in sequences, or updating the CRM, agents are better at that than you and you are behind. The things to protect are in-person time with customers and coming up with the creative campaigns agents cannot invent.
Carles’s version of it: sellers do not want to be reading 25 signals. They want prioritization, and then they want to go build the relationship.
You need to test 100 things, but I only need one of those 100 things to actually work to give me another hundred million in ARR. I only need one. I don’t need five. I don’t need 20.
You need to test 100 things, but I only need one of those 100 things to actually work to give me another hundred million in ARR. I only need one. I don’t need five. I don’t need 20.
The math underneath it is a portfolio argument. Some percentage of your plan will fail no matter what you do. Customers churn, upsells land late, a market softens. If you are not running 100 experiments to find the one, then every other thing the business does has to be perfect, and it will not be.
The grants program is the proof. It could have been a large amount of free usage and nothing else. It ended up as over 10% of enterprise revenue.
5 More Learnings From Carles That Didn’t Make the Top 8:
Expansion commissions run 12 months, and in some cases 24 months or longer. Eleven Labs ran land-and-expand with AEs hunting new logos and a customer success team owning expansion. The AE still gets paid on that expansion for at least a year. Carles calls this double compensating on purpose, and it is the same logic as paying the human when the agent closes.
The 167% blended attainment comes from a deliberate mix of rep profiles, not one archetype. Some reps carry strategic deals that take nine months to close. Others work mid-market and SMB and close inside a week. Hiring only one of those two produces a much lumpier number.
Outcome-based pricing works better as a campaign than as a permanent list price. Carles’s suggestion for removing signup friction: launch on outcome pricing so there is nothing to negotiate, then lower the per-outcome price in exchange for a commitment. The discount is the upgrade path.
Sam moved Monaco off an annual fixed platform fee to usage-based pay-as-you-go. Under the old model a customer had to sign a year contract before getting started at all. The change drove more demand and less friction, and they pair it with doing as much of the onboarding work themselves as possible, which matters more the further upmarket you sell.
Carles on large deals: go camp in their offices. His position is that selling is still a relationship problem, and no amount of agent tooling signs a big contract if you have not sat with the buyer and understood their constraints. The tooling exists to buy back the hours so you can go do that.
Three Things To Copy, and One Mistake Not To Repeat
Three things from this episode port to almost any B2B company, and none of them are the 20x number everyone will quote.
#1. Write the thesis before the launch, with a scored result at 3 to 6 months, so a market entry is an experiment and not an open-ended commitment.
#2. Aim the free tier at the segment your competitor is living on, and size it so they cannot match it.
#3. Pay commission on what you report to investors, and keep paying the humans when the agents start closing. The second half of that is what determines whether your go-to-market team helps you deploy agents or quietly waits for them to fail.
-> And the mistake to avoid is the one both of them made independently at different companies in different decades: hiring go-to-market operations and enablement years after they needed it.
Eleven Labs Just Hit $330M ARR in 24 Months. It Took Twilio 8 Years to Get There.
10 Months Ago, We Were Barely Using Salesforce. Now It's Our AI Agent Hub.
Dear Saa Str: How Should I Pay Sales Reps When Our Customers Pay Monthly?
Eleven Labs Just Hit $330M ARR in 24 Months. It Took Twilio 8 Years to Get There.
10 Months Ago, We Were Barely Using Salesforce. Now It's Our AI Agent Hub.
Dear Saa Str: How Should I Pay Sales Reps When Our Customers Pay Monthly?
Get from
Key Takeaways
-
AI VC AI Mentor: Digital Jason + Amelia AI Startup Benchmarking
-
AI Agent Playbook Free e Books
e Book: Hiring a Great VP of Sales e Book: Raising Capital e Book: The First $1m ARR -
University All Posts Podcasts The Top CROs VC Fundraising Top Videos Q&A Best of Saa Str #1 Bestselling Book Search Everything Join the Community
-
Free e Books
e Book: Hiring a Great VP of Sales e Book: Raising Capital e Book: The First $1m ARR -
AI Annual 2026 Events Overview Sponsors
Event Sponsorship Media Sponsorship
