The Human Capital Flywheel

I'm going to say something that sounds like management heresy and behaves like engineering truth:

Slack is not waste. Slack is throughput.¹

In every complex system (software, operations, teams) running "fully utilized" looks efficient on paper and feels fragile in practice.² Queues build. Context-switching spikes. Surprises become crises. People stop doing prevention and start living inside reaction.

AI doesn't change that physics. But it gives leaders a new lever: it can remove large chunks of machine-facing work that humans have been stuck doing for decades. The question is what you do with the reclaimed time.

Here's the thesis we're underwriting at Sonoran Capital Investments (SCI), and the ethos I'm personally insistent on in how we think about product and company design:

Automate the sludge → create slack → protect slack → reinvest slack into compounding human work.

That's the Human Capital Flywheel.

It's humble because it doesn't promise miracles. It's confident because the alternative fails the same way every time: you keep utilization pinned, you stay reactive, and you slowly trade craft for clerical work.

Step one: automate the sludge, not the soul of the job

Most AI rollouts start in the wrong place because they start where the demo is fun.

They try to automate the hardest judgment call first.

But the early leverage in AI isn't judgment. It's paperwork translation.

In real organizations, an embarrassing amount of time is spent doing work that exists mainly to satisfy systems: summarizing calls, updating tickets, re-entering the same information across tools, hunting for context across tabs, generating follow-ups, and producing reports that reassure everyone the workflow is "under control."

AI is good at this category. Not perfect. But good enough to reduce the human burden dramatically.³

And once you remove that burden, you can finally see what your team looks like when they're doing the actual job.

Step two: protect slack like it's infrastructure

This is the step leadership sabotages without meaning to.

Automation saves time. Management sees "capacity." Management fills it. Utilization stays pinned. The team stays underwater. Then everyone wonders why morale and quality didn't improve.

If you want the flywheel, you have to do the uncomfortable thing: lower utilization targets on purpose.²

Headroom is what allows prevention instead of reaction. It's the margin that lets people create playbooks, improve onboarding, build feedback loops, and take the extra ten minutes that turns "resolution" into a real fix.

If reclaimed time becomes instant throughput, you don't get compounding. You get a faster queue.

Step three: reinvest slack into work that compounds

Once slack exists, you get to choose what kind of organization you're building.

You can optimize for volume, which is easy to measure and usually looks good in the short term.

Or you can optimize for capability, which is harder to measure but brutal in its long-run impact.

Compounding work is practical. It looks like real mentorship. It looks like eliminating recurring incident classes. It looks like simplifying workflows so they require fewer human patches. It looks like relationship building with customers and partners so fewer interactions become escalations. It looks like better judgment because people aren't rushing.

That's how human capital compounds: by investing in skills and structures that make tomorrow cheaper than today.

The anti-pattern: counting activity and calling it progress

If you want to fool yourself, measure "AI usage." Count prompts. Count summaries. Count "agents deployed."

Those are activity metrics.

If you want to know whether the flywheel is spinning, look for downstream signals: cycle time improving, repeat incidents dropping, new hires ramping faster, senior retention improving, escalations trending down, and customer trust rising over quarters, not days.

A one-time productivity bump isn't a flywheel. Compounding improvement is.

Why we think this matters now

Most organizations are still optimized for a world where humans have to babysit software. AI changes that constraint. It allows the system to do the system-facing work.

That means leadership can choose a different operating model: one where headroom is protected, mentorship actually happens, and teams get stronger over time instead of just busier.

We're early at SCI, and we're not claiming this is "proven." We are saying the physics are clear, the incentives are aligned, and the organizations that use AI to buy headroom, rather than to crank utilization, will compound.

If this resonates, reach out. We'd love to hear how you're thinking about it.

— jason