Will AI Replace Salespeople?
The Honest Answer

Straight answer first, since that's how we write these: AI will not replace salespeople whose job is trust. It will absolutely replace salespeople whose job is information. If a rep's entire value is knowing the catalog, reciting the specs, and taking the order, that rep is competing with a machine that does all three instantly, politely, at midnight, for free. If a rep's value is earning belief from a stranger with real money at stake — that job is getting more valuable, because everything around it is getting automated.

That's the whole thesis. The rest is the evidence and what to do about it.

What AI already does better than your reps

• Answering known questions. Product specs, pricing tiers, compatibility, policy details — a machine with your documentation beats a rep's memory every time, and the buyer already knows it. They asked the machine before they called you.
• Research and prep. Company background, role, likely pain points — assembled in seconds. A rep doing this manually isn't diligent anymore; they're slow.
• The admin layer. Notes, logging, scheduling, drafting — the hours that used to eat half a selling day. We covered the six uses that pay in detail.
• The first touch. Instant, useful, 24/7 response to a new lead. Speed wins deals, and machines don't sleep.

Look at that list honestly and notice what it is: it's the job description of an order-taker. That's who's getting replaced — not by a robot walking in the door, but by buyers who simply stop needing the human version of a FAQ page.

What AI can't touch — and why

A sale, at the moment it actually happens, is a transfer of confidence between two humans. The buyer isn't asking "what are the specs" — the machine answered that days ago. They're asking, silently: can I trust you with this money, this risk, this embarrassment if it goes wrong? Answering that question takes things machines don't have:

• Stakes. A rep can put their name on a promise and be accountable for it Monday. An AI has nothing at risk, and buyers can feel the difference.
• The read. The hesitation behind "sounds good." The spouse's glance. The objection that's actually a fear wearing a costume. Live human signal, read live.
• The picture. Picture-painting works because the buyer believes the painter has seen the after-state — has walked other people just like them into it. Borrowed conviction from a language model doesn't close. Lived conviction does.
• The relationship. Renewals, referrals, the book of business — the compounding side of sales runs entirely on human trust maintained over years.

The floor of the future is smaller and better paid

Here's the forecast we'd actually stake money on, because it's already visible: teams don't go to zero reps — they go to fewer reps, each running more pipeline, each worth more. The admin layer automates, the order-taking migrates to self-serve, and what remains is the concentrated human work: discovery, trust, the close, the relationship. The reps who survive that consolidation aren't the ones who resisted AI — they're the ones who drilled the fundamentals AND took the leverage. The ones who did neither are competing with software on software's terms.

What to do about it — owner's edition

1. Automate the task layer now. Notes, first-touch, research, drafting. Every hour reclaimed goes to live selling — the part that pays.
2. Drill the human layer harder than ever. Discovery, objections, the close — premeditated and rehearsed. The differentiation shrank to the live conversation; own it completely.
3. Rebuild comp around what humans do. If you're paying commission for order-taking, you're subsidizing the past. Pay for closes, retention, and expansion — the trust work.
4. Hire for the unautomatable. Coachability, live wit, honest conviction. The role-play test matters more than it ever has, because the résumé skills are the ones going to software.

The timeline question nobody answers honestly

How fast does this happen? Faster than the skeptics say, slower than the vendors say — and unevenly on purpose. Transactional selling automates first, because the trust required is low and the volume is high; you can watch it happening in real time wherever buyers already prefer self-serve. Complex, high-stakes, relationship-driven selling automates last, because the bottleneck was never information — it was confidence, and confidence doesn't ship in a model update. The pattern matches every automation wave before it: the technology arrives years before the buyer behavior does, and the operators who win are the ones who move with the buyer's comfort level instead of ahead of it or behind it.

Which means the two failure modes are both timing errors. Panic-automating your whole funnel this quarter hands your buyers to a competitor with humans on the phone. Ignoring it for five years hands your margins to a competitor with half your payroll. The boring, correct play: automate the task layer continuously, re-audit the seat map yearly, and let your own close-rate data — not a keynote — set the pace.

Straight answers

Tino LardiCo-founder. Built the lead engine behind a ~$1.3M-a-year recurring book. MC Mike CatoggioCo-founder. Thirty years of selling, drilled into a method.