One of the most common questions many of us are wrestling with now: just how many reps do I actually need now?
And there are two very different playbooks being run right now across the fastest-growing AI companies. Both work. Both have real trade-offs. And the one you pick often comes down to who’s running your sales org, and how much capital you’ve raised.
Let me break them down.
Playbook #1: The Mega Quota — Just Service the Massive Inbound
Many of the hottest AI B2B companies are running this playbook today, and the numbers look nothing like traditional SaaS.
Often up to $4M+ quotas per AE.
That sounds absurd if you’re used to the old world. But here’s the math:
When you’re the hot vendor in a white-hot category, your inbound leads are extremely high intent. These aren’t tire-kickers. These are buyers who’ve already decided they need what you sell — they just need to pick a vendor. And you’re the one they want.
So a rep can realistically close 2x as many deals as they would in a normal environment. If the average deal is ~$50K, and a rep is closing 10 deals a month, that’s:
10 deals x $50K = $500K/month = $6M/year in bookings
That’s very doable at the companies with the strongest product-market fit right now. When demand outstrips your ability to service it, reps don’t need to hunt. They need to execute.
- The upside: Super capital-efficient. Small team, massive output per head. Your sales org stays lean, margins stay strong, and you avoid the chaos of scaling too fast.
- The downside: You leave a lot on the table. You certainly can’t service every potential deal.
When your team is small and quotas are sky-high, reps naturally focus on the hottest, highest-intent leads. The ones that are basically pre-closed. And that’s rational — there’s more than enough demand to hit plan.
But all those “pretty good” leads? The mid-market prospects who need a second call? The enterprise buyer who wants a custom demo? They never get a callback.
These are the companies where prospects complain they literally can’t get a meeting. You’ve probably experienced this yourself. The hottest AI vendors right now are just hard to even book a call with. That’s a feature of Playbook #1 — but it’s also a bug.
You’re hitting your numbers. But you’re not capturing anywhere close to the full market opportunity.
The ElevenLabs Case Study: A Hybrid, But Still ~$2M Per AE
ElevenLabs might be the most interesting example of how a hot AI company is actually doing this.
Carles Reina, their VP of Sales (who was also the first investor and fourth employee), just did a deep dive on 20VC that’s worth studying. He scaled the revenue org from Day 1 to over $330M ARR in roughly 3 years — with a team that’s still relatively small for that revenue level. The company has around 500-700 employees total, not thousands.
The headline number everyone grabbed: ElevenLabs sets quotas at 20x base salary. If your base is $100K, your quota is $2M. And they’re ruthless about it — miss quota, and you’re out.
That sounds brutal, but here’s what makes it work: more than 80% of their reps hit quota. That’s not a quota designed to punish. It’s a quota designed to filter for elite performers and then set them up to succeed.
A few things from their playbook that are directly relevant to this debate:
- They started at 90% inbound — and deliberately pivoted to 50/50 inbound/outbound. This is the insight most hot AI companies miss. Reina recognized that pure inbound dependency is a trap. It feels great while demand is surging, but it doesn’t build the muscle you need for the next phase. So even while inbound was crushing it, they invested in outbound to build long-term pipeline health.
- They land small and expand aggressively. Deals often start at $12K and grow into millions. Both the AE and the CSM get paid on upsells — double comp on expansion revenue. Reina’s logic: paying double on expansion is worth it because you get two people working their hardest to grow every account. That’s the land-and-expand motion working at its best.
- Reps should be on the road, not in the office. Reina is emphatic that sales reps sitting in an office doing virtual meetings are wasting the company’s money. Get in front of customers. Get on planes. The face-to-face close rate advantage is massive — Toast’s data backs this up too, showing prospects who get an on-site visit close at 3x the rate of those who don’t.
- They forecast pessimistically on purpose. Always underestimate deal sizes. Always assume deals will slip. This forces the team to build a bigger pipeline than they think they need, which prevents the feast-or-famine cycle that kills so many sales orgs.
The ElevenLabs model is essentially a hybrid: the per-rep productivity of Playbook #1, but with the outbound discipline and pipeline coverage of Playbook #2. Small team, high output per head, but nobody’s waiting around for inbound to ring the bell.
Playbook #2: The Traditional Quota — Service Every Lead, Hire Like Crazy
This is what tends to happen when a hot AI B2B startup crosses $50M ARR and brings in a seasoned CRO from the pre-AI world.
The new CRO looks at the plan. Let’s say the board wants $150M in new bookings next year. They run the math the way they’ve always run it:
$150M in bookings / $700K quota = ~200 reps needed. Tomorrow.
And suddenly you’re in hypergrowth hiring mode.
This isn’t wrong, exactly. It’s just a completely different philosophy. The CRO’s logic is: every lead should get a call back. Every prospect should get a demo. Every opportunity should be worked.
- The upside: Nothing falls through the cracks. Every prospect gets touched. You maximize coverage across the full funnel, and you’re building the kind of sales machine that can sustain growth at scale.
- The downside: Chaos.
When you go from 20 reps to 200 in a year, you outrun your infrastructure. There aren’t enough SEs to support all those AEs. There aren’t enough SDRs to properly qualify inbound. Onboarding breaks. Enablement can’t keep up. CRM hygiene goes out the window.
And the per-rep economics look a lot less impressive. You’re trading efficiency for coverage — which can be the right trade, but it’s expensive and messy.
The Broader Trend: AI B2B Sales Teams Are Running at Half the Headcount
Zoom out from these individual playbooks and the macro picture is clear: AI B2B companies are running sales teams that are roughly half the size of their predecessors 24 months ago.
The data from ICONIQ’s GTM benchmarking backs this up. The percentage of budget going to sales is staying roughly flat — around 55-56% — but the absolute headcount is shrinking dramatically. AI-native companies are achieving close rates 50% higher than traditional B2B, which mathematically means you can get by with fewer reps.
Just look at the revenue-per-employee numbers across AI leaders right now. They’re staggering:
- Anthropic: Running at $9B+ annualized revenue with ~2,000-3,000 employees. They have less than 100 quota-carrying sales reps. For a company doing $14 Billion+ ARR. 80% of revenue comes from API customers and enterprise accounts, many of which are largely self-serve or partner-driven (AWS, GCP). Their GTM org is scaling 3-4x year-over-year per their own job postings, but the starting base is tiny.
- OpenAI: $20B+ revenue with ~4,000 employees at end of 2025. Only 58 quota-carrying sales reps for the entire company. That’s ~$345M in revenue per sales rep.
- Cursor: Hit $2B+ annualized revenue as of March 2026 — doubling in roughly 90 days — with just over 300 employees. Enterprise customers now account for ~60% of revenue. They were only building an enterprise sales team in early 2025 — 4,000-5,000 companies reached out requesting enterprise access before Cursor even had a team to take the calls. They hit $100M ARR as the fastest SaaS company ever, with zero marketing spend. That’s ~$6M+ ARR per employee, and accelerating.
- ElevenLabs: ~$330M ARR with 500-700 employees. Started 90% inbound, now 50/50. Small sales team, 20x base quotas, 80%+ attainment. $500K-$660K revenue per head.
- Harvey: ~$190M ARR with ~860 employees — more traditional, but they invested heavily in lawyers and domain experts building the product, not just AEs. Grew from 5 employees to 340 in under 3 years.
Cathy Gao at Sapphire Ventures put it well: companies are now scaling to $60M in ARR with 30 employees. Fathom AI’s CEO has said their internal goal is $100M revenue with fewer than 150 employees — and with AI tools, that’s gone from insane to doable.
In the 2000s, it took 500-1,500 employees to hit $100M ARR. In the PLG era, it was 300-500. Now? Some AI companies are doing it with fewer than 100 people.
And it’s not just the quota math that’s changing. Founders and VPs of Sales are asking harder questions:
- Is headcount better allocated to FDEs (forward-deployed engineers) than more AEs?
- Are human SDRs worth it when inbound demand is this strong and AI tools can handle qualification?
- If AI agents can do the first demo, the qualification call, and the follow-up — what exactly do you need 200 humans for?
The “cracked” rep — the elite performer who can handle complex enterprise deals and earn $600K+ — is more valuable than ever. The mid-pack inside sales rep working from home? That role is disappearing fast as AI agents take over the routine motions.
So Which Playbook Wins?
I think most founders instinctively prefer Playbook #1. The Mega Quota. I do myself.
It feels cleaner. It’s more capital-efficient. It lets you stay focused. And when you’re the hot company, it works beautifully. Your reps are crushing quota, your cost of sale is low, and you don’t have the headaches that come with managing a massive sales org.
But here’s what happens in practice:
You raise $100M+ and bring in a seasoned management team. And that team — the CRO, the CFO, the board — they see all the demand you’re not capturing. They see the leads going unanswered. They see competitors starting to pick off the prospects you’re too busy to call back. And the pressure to go big becomes enormous.
The board didn’t give you $100M to run a boutique sales team.
So many of the hottest AI companies end up migrating from Playbook #1 to Playbook #2 as they scale. Not because the founders wanted to. But because the math — and the pressure — demands it.
ElevenLabs is interesting because they’ve found a middle path: stay lean, keep quotas sky-high, but build real outbound muscle so you’re not at the mercy of inbound alone. Reina’s philosophy — high output per head, ruthless accountability, and reps on planes not in offices — might be the template more AI B2B companies should study.
Are We All Getting More Efficient Anyway
The reality even non-AI B2B companies maybe converging on similar modals. They are beginning to drive quotas up, because the VC funding and cash just isn’t there to overhire. And that may mean sacrificing growth for many. But they may have no choice.
If Your Lean GTM Team is Crushing It — Maybe Let It Ride Longer Than The “Experts” Tell You
What I’d tell most AI B2B founders right now: you can run leaner on the sales team than almost anyone will advise you to.
As long as the inbound demand is intense, growing, and — this is the important part — often pre-qualified. Many of your best prospects are already using the product. They’re on a waitlist. They’ve been in your free tier for months and they’re ready to swipe the card. When that’s the dynamic, you don’t need 200 reps. You need 20 great ones.
Yes, bring in that experienced CRO eventually. But let the lean model ride a bit longer than the board or your new VPs might tell you.
Because the headaches that come with 10x-ing the size of your GTM team are real and they are brutal. It becomes near impossible to keep the quality bar up. You’re onboarding reps faster than you can train them. Half of them don’t understand the product well enough to sell it properly. Pipeline reviews go from tight and honest to performative. The culture that got you to $50M ARR starts to dissolve.
But the bigger thing — and I think this is the one most founders are still missing:
Most Importantly. In the AI age, Sales is Often Not the Real Bottleneck Anymore
In traditional B2B SLG, it was. You needed reps to generate demand, qualify it, and close it. The sales team was the constraint on growth.
But with AI B2B products today? Especially agentic products? The bottleneck has shifted to deployment. FDEs, onboarding specialists, AI agent training, getting the product actually working inside a customer’s environment — that’s where deals die now.
Highly trained FDEs and onboarding resources may matter more than your next 10 AEs. Almost none of these hyper-scaling AI Sales teams I’ve worked with have enough FDEs and other resources to properly deploy their products.
If you can close them but you can’t support them, it’s actually better not to close them at all. A customer who signs a $200K deal and then churns in 6 months because the agent never got properly deployed is worse than a customer you never sold. They’re a detractor. They’re a negative reference. They’re wasted FDE time you’ll never get back.
The worst possible outcome? Massively scaling up the sales team while starving the team that actually gets the AI agent and product deployed. You’re writing checks your implementation org can’t cash.
I’d rather reverse it. Scale the deployment and onboarding team first. Make sure every customer you close gets white-glove treatment and actually goes live. Then — and only then — turn up the sales volume.
The AI B2B companies that win long-term won’t be the ones with the biggest sales teams. They’ll be the ones where every customer they close actually succeeds.
