The "1.2 Human" Sales Team is Here (And It’s Mostly Just Better Ops)
The one thing you need to know in AI today | AI Ready CMO
If you want a preview of what the next 18 months of B2B go-to-market look like, veteran SaaS investor Jason Lemkin just provided an incredible blueprint. The headline: Lemkin replaced a 10-person sales team with 1.2 humans (yes, 1.2) and 20 AI agents. Revenue stayed flat.
Look past the shock value. This isn’t a sad story about robots coming for your job. It’s about the death of the “mediocre middle” in knowledge work, and we all should celebrate that.
For years, we’ve accepted a bloated model in sales where junior employees burn out doing robotic tasks—qualifying inbound leads, sending generic cold emails, chasing dead opportunities. Most of this work can be broken down into repeatable workflows, discrete steps you can map on a whiteboard. Even the judgment calls like “is this lead worth my time?” can be automated with good reliability if you train the system on your best people’s instincts. And AI doesn’t get tired, doesn’t complain about “trash leads,” and works Christmas morning.
But there are things that only human beings do well. AI can help you prospect, triage, and qualify much better than the best SDR.
But it can’t build trust.
This is especially critical for high-ticket, complex deals, where you need a human closer (the “1.0” in his equation). For anything inexpensive, buyers actually prefer the speed of an agent over waiting two days for a “relationship builder” to call them back and ask about the budget.
The most interesting role in this new setup isn't the closer, though. It's the "0.2" called Amelia. She's not a salesperson, she doesn’t even have a sales background. She's a Chief AI Officer, which in practice means a hyper-competent ops lead who spends 10-15 hours a week orchestrating the agents: checking for hallucinations, resolving conflicts between bots, updating training data.
Lemkin’s agent army didn’t work out of the box. The first agent started sending customers wrong event dates, and he spent an hour every morning fixing its mistakes. But after training, that same agent closed a $70K sponsorship deal on its own. It took him months to perfect the system, so his learnings are worth considering:
Expect that it won’t work at first. The biggest lie in AI sales is that you can just “turn it on.” You can’t. Someone needs to review the agent’s outputs daily, correcting mistakes until it learns your context. Skip this step and you’ll get nothing.
Don’t build, buy. Unless you’re an engineering-first company, your internal tool will be obsolete in three months. Buy established tools that come with “Forward Deployed Engineers”—vendor staff who ensure the agent actually works before you pay.
Start with the trash. Don’t put an untrained agent on your VIP leads. Deploy it on the leads your team ignores: small accounts, ancient opportunities, “Contact Us” spam. Lemkin’s agents hit a 70% response rate on “dead” leads humans deemed worthless. Found money, zero risk.
Train on your best, not your average. Upload your top rep’s emails, scripts, objection handling. The agent becomes a scalable copy of your best performer. Feed it mediocre inputs, get mediocre outputs at scale.
Your orchestrator is an ops lead, not a salesperson. The Amelia role requires someone who loves data and understands logic flows—not a traditional sales manager. If you hire a “people person” for this job, they’ll try to motivate the software instead of debugging it (which is a fascinating insight).
Most people will focus on the headcount reduction. That’s fair—it’s real and it matters. But let’s be honest: no one loved the mediocre middle. Not the people stuck doing it, not the managers overseeing it. We kept these roles because we had no better option. Now we do.
3 AI Marketing Tools To Try Today
Gamma
Create polished presentations in minutes without design skills—Gamma AI generates slides, formats content, and suggests layouts automatically, turning outlines into professional decks instantly.
TextCortex
Write anything faster with AI that adapts to your style—TextCortex generates emails, documents, and social posts while learning your voice and preferences over time.
MeetAlfred
Automate LinkedIn outreach and multi-channel campaigns—MeetAlfred manages connection requests, follow-ups, and email sequences across platforms, turning cold prospects into warm leads without manual work.