Buyers still want a human. What should your sales team change?

Your buyers do most of their research with AI now, and most of them still want a person to check what it told them before they commit. For a startup's sales team, that changes the job: you'll meet buyers later, they'll know more, and what they need from you is help confirming, correcting and de-risking a decision they've mostly worked through on their own.

I wrote the first version of this on Substack in July. The numbers have held up, and some newer ones make the case stronger.

What buyers are telling us

In May, Gartner published a survey of 645 B2B buyers, run in August and September 2025. Sixty-nine percent said they prefer to validate AI-generated insights with a sales rep. Buyers used an average of seven information sources on a recent purchase, and 45% said they'd used GenAI, mostly to gather information on vendors and products.

Gartner also reports that 67% of buyers prefer a rep-free experience. That looks like a contradiction until you see where reps still win. In the May survey, reps were the most important information source when buyers were researching a business problem, identifying a preferred supplier, securing internal support and finalizing the purchase. Buyers want to do the reading themselves. They want a person for the decisions.

The longer view points the same way. Gartner predicted in August 2025 that by 2030, 75% of B2B buyers will prefer sales experiences that prioritize human interaction over AI. Its analysts expect the pull back toward people to be strongest in complex, high-stakes deals.

Since my original post, G2's 2026 Buyer Behavior Report surveyed more than 1,000 software buyers and found that 47% would let an AI agent research and recommend while people keep the final decision. Finance involvement in software decisions also rose from 31% to 46% in a year. More of the reading is going to machines, and more people want a say in the decision.

The floor went up, and trust got scarcer

AI raises the floor, and the value moves to the ceiling. That's the core of the AI chapter in my book, Curiosity. Research, first drafts, call notes and follow-ups used to separate a good rep from a poor one. Now your laziest competitor has them too, for nearly nothing, so the difference sits in the parts a machine can't do: curiosity, judgment, honesty and trust.

The most sobering figure in Gartner's survey is the one about trust. Fifty-one percent of buyers said they're more likely to get misleading information from GenAI, and 49% said they're more likely to get it from a sales rep. When I first wrote about this I was surprised the AI number was that low. The rep number is the one to worry about. Buyers come to us to check the machine, and almost half of them think we're the more likely to mislead them.

AI does get things wrong. It has handed me confident facts about companies that turned out to be wrong, which is why I check anything I'd repeat to a customer. A rep who's desperate for the deal gets things wrong too, and buyers have had longer to learn that. If a person is going to be the check on the machine, they have to be more trustworthy than the machine. That's a higher bar than most sales teams are set up for.

What I'd change in a startup sales team

Gartner's own advice is that sellers "need to show up differently" at the moments where they help buyers validate information and reduce risk. In practice, for a team of two to twenty reps, I'd do this:

• Assume they've been briefed, and ask what they were told. Most buyers arrive with an AI's summary of you and your competitors that you didn't write. Ask what they've read and what they believe. Confirm what's right, correct what isn't, and own the parts that are right about your weaknesses. (Related: your buyers' AI agents are reading your website.)
• Lead with their problem. Gartner found reps are the most important source while buyers are still working out the business problem. Spend the first call on their business, and hold the demo until you know which part of the product is worth showing. (More on that in what "adding value" means when buyers ask AI first.)
• Make every claim checkable. Show the product doing the specific thing they asked about. Offer a customer they can call. Put the limits in writing. A straight "no, it doesn't do that" early on is the fastest way to be believed on everything else.
• Help your champion get it approved. With finance in nearly half of software decisions, your champion needs a business case in their own numbers that a CFO can read in five minutes. Write it with them.
• Give AI the scaffolding and keep the judgment. Gartner predicts AI agents will outnumber sellers 10 to 1 by 2028, yet fewer than 40% of sellers will say agents improved their productivity. Use AI for research, call notes, CRM updates and follow-ups so your reps get hours back for buyers. Don't use it to send more email. I get a steady stream of AI-written outreach, and the giveaway is how often several messages reference the same post I wrote, or a comment I left somewhere, in nearly identical words. (Getting this balance right is what our AI for revenue teams work is for.)
• Hire for curiosity over polish. Gartner's 2030 prediction came with advice to build fewer, more specialized sales roles and invest more in interpersonal skills. For a startup, that means your first sales hires should be the people who ask the best questions and tell the truth when it costs them, even if someone else gives a slicker demo.

Keep the founder in the room

If buyers want a person to check the decision, early buyers want the person with the most at stake. That's usually the founder. Even with a sales leader in place, I'd keep founders in the first few dozen deals, at least for the moments where a buyer needs to hear "yes, we'll stand behind that" from someone who can say it.

Where founders get stuck is doing all of it themselves: the research, the demos, the follow-ups, the proposals and the forecast. That's where a fractional CRO helps. We build the team and the process around you so your time goes to the conversations buyers want you in.

The buyer's AI won't pick up the phone when the implementation goes sideways. Make sure your rep is someone they'd want to call.