> ## Content Index
> Fetch the complete content index at: https://www.kerrest.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The Comeback of the Humans
- URL: https://www.kerrest.com/2027-comeback-of-the-humans/
- Published: 2026-11-13T18:30:00.000Z
- Updated: 2026-11-13T19:51:29.000Z
- Description: The more AI does, the more building and running a company comes down to people.
- Author: Frederic Kerrest
- Tags: Essays, FK’s 2027 Predictions

Two years ago, the consensus was clear. AI was going to end sales. Agents would buy from agents. Startups would become one-person, billion-dollar companies. And the MBA? Worthless.

Fast forward to today. The number one question I get from founders isn't about models or GPUs. It's: how do I go from three pilot customers and $1 million of founder-led sales to dozens of customers and $10 to $100 million (and then $1 billion!), as fast as possible? The answer is always the same: start building your enterprise sales force, yesterday.

Welcome to the comeback of the humans.

## The jobs apocalypse that didn't happen

If you followed the coverage last spring, you'd think AI was about to wipe out the workforce. The truth is, it hasn't.

Look outside tech. Big, non-tech companies are not laying off thousands of people because of AI. Unemployment is 4.1%. Job cuts that employers actually blamed on AI peaked in May at about 38,600, then fell to about 1,300 in October, the lowest of the year.

So what about all those tech layoffs? Mostly, that's a COVID hangover. Between 2019 and 2022, Amazon nearly doubled its headcount. So did Meta. Alphabet grew 60%, Microsoft 53%. Money was free, everybody worked from anywhere, and companies hired like it would never end. Back in 2022, Stripe's CEO said it plainly: they had overhired. Now a lot of companies are cleaning that up, and AI makes a convenient, stock-friendly explanation.

Even Microsoft's buyout offer this fall, open to about 60,000 people, excluded sales, marketing and data-center roles. Draw your own conclusions about which jobs they think matter.

And while everyone was predicting the end of software jobs, Indeed's software postings climbed about 15% off their low, even as overall postings fell.

## Enterprise AI moves at enterprise speed

Here's what happens inside a big company when it decides to "do AI." It's not a demo. It's process, systems and data. Who owns the workflow? Where does the data live? What happens when the model is confidently wrong?

That's why adoption is slower than the hype. Only about 17 to 20% of US firms use AI at all. An MIT study found only about 5% of enterprise generative AI pilots produced fast revenue gains. A lot of early deals were one-year contracts that the CIO bought because the board asked, "What's our AI strategy?" Then nobody deployed them.

And the prize is enormous. Sequoia sizes the services market AI is going after at about $10 trillion, against roughly $650 billion for software. Rewiring that much of the economy won't take two years. It'll take a generation. And somebody has to manage every step.

## Sales is back

Sure, in small businesses, agents may really end up buying from agents. Jason Lemkin replaced his SaaStr sales team with 20 AI agents run by "1.2 humans." More power to him.

On January 31, 2015, the last day of Okta's fiscal year, I got on a last-minute flight to Phoenix. We needed one more deal to make our number, and it was with American Express. No email or demo was going to close it. So I sat across the table from their vice president of IT, and we signed it that night. That's still how the enterprise buys, and the enterprise is where most software, and most AI, gets bought. People buy from people.

At 515 Ventures and Windproof Partners, I see the same thing every week. Founders with great products and no idea how to sell them. The hottest conversations are about who to hire, who knows how to sell, and who the sales leaders are. We're right back where we started.

The numbers have changed, sure. Top enterprise reps now carry quotas of $2.5 million or more, up from about $1.5 million. But you still need a lot of them. And you need the whole stack: SDRs and BDRs (maybe with some machines helping on lead gen), account executives, line managers, the process, MEDDPICC, enablement. Nobody becomes a great salesperson without making a few hundred cold calls first.

Sales has been undervalued for twenty years as more product and engineering founders took over. It's an art and a science. It's a career. And it's back.

## Culture, management and the MBA

Here's my personal definition of culture: it's how people behave when no one else is watching.

Now think about agents. More than half the time, no one is watching the agents.

We've already seen what happens. A coding agent deleted SaaStr's production database during a code freeze, then invented about 4,000 fake records. In a test, one of OpenAI's models rewrote its own shutdown script so it wouldn't be turned off. Agents can swarm, hack, mislead and take things they shouldn't, because they're chasing a goal.

So the job is changing. Managers are going to run teams of human and digital workers. Somebody has to set the culture, the guardrails and the incentives for both. Somebody has to enable both. That's a whole new discipline, and it will evolve for the next decade or two.

Which brings me to the MBA. Applications to US full-time MBA programs fell 27% this year, mostly because international students are going elsewhere. I get it. It's expensive, and people question the value.

Here's my controversial take: the MBA is about to matter more, not less. Business administration now means administering humans and machines. Building teams. Building culture. Building consensus. Presenting. Handling objections. And the network of people who learn those things together matters as much as ever. (Full disclosure: I have an MBA from MIT Sloan, I chair its visiting committee, and I teach MBA students at Berkeley Haas. So take that with a grain of salt. I'd say it anyway.)

## My prediction for 2027

In 2027, the scarcest hires in AI won't be researchers. They'll be enterprise sales leaders, and managers who know how to run teams of people and agents.

So if you're building a company, hire a great sales leader earlier than feels comfortable. If you're a manager, start learning how to lead a mixed team now. And if someone tells you the humans are done, don't believe them.

We're just getting started.

*This is the third of my five predictions for 2027\.* [*Read the full series.*](https://www.kerrest.com/tag/2027-predictions/)