"This time it's different" is what people always say when stocks are rising. It’s what they say about AI today.

Frank walked me through the pattern: In the 1930s, Fortune magazine ran articles predicting that telephones would eliminate salespeople. “Once we have phones throughout the United States,” the thinking went, “buyers and sellers can call each other directly and negotiate price”. In the 1950s, the national highway system would let customers drive around and shop for the best price themselves. In the 1990s, the "information superhighway" promised the same revolution.

None of it happened. Each technology made a real difference—the telephone accelerated productivity, changed buying patterns, and altered sales tasks. But it didn't eliminate the salesperson. AI will follow the same arc.

Here's the data point that matters: salespeople spend only 30-35% of their time in actual customer contact. The rest goes to administrative tasks, CRM updates, and report writing.

Frank argues that AI excels at exactly this work. Email cadences, CRM inputs, customer research—AI handles these faster and often better than humans. It’s not just a productivity gain. When you save time in sales, you expand your total addressable market. Segments that were economically infeasible suddenly become viable.

Frank talked to me about a case study he worked on with SAP: They had avoided the small and medium enterprise (SME) segment for years because going after it was simply too expensive.

AI changed that. SAP compressed their sales cycle from nine months to nine weeks. They reduced costs, increased reach, and most importantly, redesigned the sales role itself: Salespeople in the SME segment weren't closing deals in the traditional sense. They were coordinating—acting as the interface between the customer and SAP's product and service teams, making sure the solution fit.

That's the real lesson. The technology enabled market expansion, but only because SAP rebuilt their entire sales organization around it.

You can buy a $1,000 iPhone without ever talking to a salesperson. The entire transaction happens through machines. That should be the future, right? AI handling even high-value purchases?

But even Apple—the company that arguably needs salespeople the least—built physical stores. When Steve Jobs was considering this move, many thought it was unnecessary. They were wrong. The stores worked precisely because they provided a curating function that self-service couldn't match.

Frank argued that people establish trust with other people, not algorithms. In B2B sales, especially, buyers are betting chunks of their careers on whether you can deliver. That's not a transaction a chatbot handles well.

There's a concern I hear constantly: AI is creating clutter. Everyone now assumes your email was written by AI and trust erodes.

Frank wasn't particularly worried. We've been experiencing this with digital marketing for a decade. Email marketing tools like Mailchimp already let anyone send thousands of messages. It became cluttered and expensive, but instead of disappearing, Marketing simply evolved.

AI will accelerate the clutter. Because it's so easy to use, everyone will send another 10,000 emails. That commoditizes the channel. The job of sales managers is to track when this happens and move to the next uncrowded space. That's competition.

What was the "next thing" in the 1990s—targeted email, digital campaigns—is now standard practice. AI adoption in sales will follow the same path, Frank argued. In a few years, AI won’t be a differentiator, it will be table stakes.

Companies are hiring fewer people overall. Two years ago, consulting firms and banks hired ten candidates from Harvard. Now they may hire six. But those cuts seem concentrated where AI replaces analytical tasks. Sales is a different animal.

In sales roles, companies are still hiring for sales expertise—not AI fluency. Frank was blunt about this. "Hiring someone in sales for AI is like saying 40 years ago, 'I run a telemarketing operation. I better hire people who know how to use the phone.' Doing the prompts is not that difficult. We can train for that in an afternoon. Doing selling and relationships—that's different."

I pushed Frank on this. Clay, a company whose tool I work with extensively, talks about a "go-to-market engineer"—a role that's half sales, half software engineer. Wouldn't that be the future?

Frank wasn't buying it. When you can find a Michelangelo or Leonardo da Vinci who can do multiple disciplines superbly, hire them. But there aren't many. For reasonably intelligent people—not superheroes—division of labor works better than trying to create unicorns.

I closed by asking Frank what people will look back on and regret about 2026.

His answer: confusing the technology with the business. AI is a powerful tool if you have a relevant sales model. But AI is not the same as a sales model. Many people think it is right now. They're going to pay for this belief.

The second issue is data quality. AI is only as good as your inputs. Most CRM data is notoriously noisy—not because of the software, but because of human inputs. One rep's "lead" is someone they bumped into on the street. Another's "lead" has an actual budget.

Cleaning up data to make AI useful is expensive and difficult. With large language models, it's harder still—deleting or updating datasets is more complex than in traditional IT systems. Companies are only now realizing the scale of this task.

Garbage in, garbage out. Frank says it's still true.

Follow me on Linkedin | Kintro