The Next Employee Won’t Be Human

For most of my career building companies and investing in startups, growth has almost always eventually meant adding people.

You find an opportunity, allocate some capital, build a team and start hiring across product, engineering, sales, marketing, finance and operations. As the business grows, the organization grows with it, usually bringing more management layers, more software, more processes and eventually a lot more complexity.

That relationship between growth and headcount has been so consistent that we rarely question it.

I think AI is starting to break that assumption.

The next employee your company adds may not be human. It might be an AI agent that researches a market, analyzes competitors, qualifies prospects, writes software, prepares financial analysis or coordinates work across multiple systems.

The more interesting question isn't whether AI agents will replace certain jobs. It's what happens to the economics of building and growing companies when adding productive capability and capacity no longer necessarily means adding people.

For growth innovation, I think the implications could be significant.

For most of business history, adding organizational capability usually meant adding people. If you wanted to sell more, build more, analyze more or enter a new market, you hired people with those skills.

Technology made those people more productive, but the underlying relationship between capability and headcount remained largely intact.

Salesforce helps salespeople manage customers. Excel helps analysts work with information. Slack helps teams communicate. ERP systems help large companies spend several years and millions of dollars trying to understand their own business processes.

Joking aside, AI starts to break that relationship because increasingly the technology isn't simply helping someone do the work. It can take responsibility for parts of the work itself.

This creates an important distinction between capability and capacity.

Capability is what the organization can do. Capacity is how much of it the organization can do.

AI changes both.

A small team can access capabilities across research, software development, design, analytics, marketing and operations without hiring a specialist for every function, while simultaneously increasing the capacity of the people already inside the organization.

Companies can increasingly add capability without adding equivalent headcount, while simultaneously increasing the capacity of the people they already have.

The traditional equation starts to change:

Old model: Growth → More Headcount → More Capacity

AI native model: Growth → More Capability → Humans + Agents → More Capacity

Capacity is about doing more.

Capability is about being able to do something new.

That is where AI gets particularly interesting for growth innovation.

One metric I think we're going to hear a lot more about is revenue per employee.

The ratios vary enormously by industry and business model, but historically growing a substantial business generally meant growing the organization required to operate it.

AI native companies are starting to test how far those two curves can separate.

We are already seeing private AI companies reach hundreds of millions, and in some cases billions, in annualized revenue with organizations dramatically smaller than we would historically associate with businesses of that scale.

Revenue per employee isn't a perfect measure. AI companies also spend enormous amounts on compute and infrastructure, so high revenue per employee certainly doesn't mean high profit per employee.

The signal is still important.

What happens when revenue can grow dramatically faster than headcount?

The objective isn't to build companies with as few humans as possible. It's to understand how much economic output an exceptional group of humans can create when surrounded by increasingly capable machines.

If a new venture can reach meaningful validation and revenue with five people instead of twenty, the amount of capital required to test that growth opportunity changes dramatically.

In my previous CIO in Beta post, The Best AI Strategy Isn't About AI, I wrote about thinking about AI investment across three areas: optimize the core, transform the core and build beyond the core.

Most companies are understandably starting with optimization. There is significant ROI available from making employees more productive, automating repetitive work and reducing the cost of existing processes.

I continue to think the more interesting opportunity comes when companies ask what AI makes possible that wasn't economically viable before.

Historically, testing a meaningful new growth opportunity inside a large company could become expensive surprisingly quickly. Before you knew whether customers actually cared, you could easily have a significant team and millions of dollars committed.

AI is starting to compress that process.

Small entrepreneurial teams working with specialized AI agents can increasingly conduct research, develop prototypes, analyze markets, write software and test customer acquisition without building a large team around every experiment.

When the economics of experimentation change, the portfolio of experiments a company should be willing to run should change with it.

Instead of simply using AI to make existing operations cheaper, companies should be asking whether cheaper experimentation allows them to take more shots on goal.

Large companies should theoretically have an enormous advantage.

They already have customers, distribution, proprietary data, trusted brands, industry expertise, balance sheets and relationships that startups often spend years trying to build. Startups have traditionally had something equally important: speed, focus and the freedom to challenge existing assumptions.

The problem is that putting a startup inside the operating model of a large company usually doesn't produce a startup.

It produces a small corporate project.

This is why I believe companies need dedicated teams operating outside the core with a different mandate, governance model and way of working.

Whether you call it a venture studio, venture building or something else matters less than creating an operating environment designed specifically for discovering and building new growth. These teams should identify opportunities, validate assumptions quickly, build with customers and progressively allocate more capital as evidence increases.

AI makes this model considerably more powerful.

Imagine taking valuable corporate assets and putting a small entrepreneurial team around them, supported by AI agents across research, software, customer insights, go-to-market and operations.

The objective isn't to recreate the corporate organization around every opportunity. It's to create the smallest possible team with enough capability to determine whether an opportunity deserves more capital.

The combination becomes powerful:

Corporate Assets + Dedicated Venture Team + AI Capability = Startup Speed at Corporate Scale

Every CEO, board and investment committee has limited resources and an almost unlimited number of places they could deploy them.

Historically, adding capability usually meant hiring people, buying technology, acquiring another company or engaging an external partner. AI introduces another option.

Should the next dollar go toward another employee inside the existing organization? Another technology platform? An acquisition? Or a small AI native venture team testing an entirely new source of growth?

I think the companies that figure this out early will develop a structural advantage. They won't necessarily have the biggest AI budgets or the most employees using copilots. They'll get better at turning corporate assets into new enterprise value because they can run more experiments, learn faster, kill bad ideas earlier and put more capital behind the opportunities that demonstrate traction.

That starts to look less like traditional corporate innovation and much more like venture capital inside the enterprise.

There is also a fascinating future of finance implication as AI agents begin performing economic work and eventually need budgets, permissions and the ability to transact through programmable financial infrastructure.

That's probably a topic for another post.

I don't think the future is simply companies replacing people with agents.

The more interesting future is one where organizations become deliberate about what humans should do, what machines should do and how that combination changes both the capability and capacity of the organization.

For growth innovation, the implications could be enormous.

Cheaper intelligence shouldn't just make existing companies more efficient. It should make them more ambitious about creating new growth.

The question for CEOs, boards and capital allocators shouldn't simply be how many jobs AI might replace.

It should be: What are we capable of building now that we couldn't economically build before?

The biggest economic impact of AI may not come from replacing the next employee.

It may come from realizing that you no longer need to hire the next twenty people before you can discover whether the next business is worth building.