Pointing Humans at Humans Again

I don't have a religion about AI. I have a preference for systems that don't waste people.

If you zoom out on the last few decades of "digital transformation," something pretty strange happened. We didn't just adopt software; we reorganized work so that humans spend a huge amount of time making software feel coherent. We trained smart people to become translators between messy reality and rigid systems.

We normalized it. We even called it efficiency.¹

My view (and the thesis we're underwriting at Sonoran Capital Investments, SCI) is forward-looking and simple:

The next decade of operational software will be defined by one inversion: machines will increasingly do machine-facing work, so humans can return to human-facing work.

Not as nostalgia. As a correction.

The quiet invention of the internet era: humans as adapters

The internet era delivered real progress. But it also quietly created a job category that shouldn't exist at the scale it does: the human adapter.

You can spot it by the kinds of sentences organizations treat as normal:

"Can you update the ticket so the dashboard is accurate?"
"Just copy the details into the other system, this one doesn't sync."
"Make sure you log the call or finance can't close the month."

None of that is the work. That's the paperwork we do about the work.

And the hidden damage isn't just time; it's skill drift. People get good at what their day requires. If their day requires tool choreography, they get good at tool choreography. If their day requires judgment, mentorship, and relationship work, they get good at those. You can guess which set compounds.

The shift we care about: from "forms first" to "conversation first"

Most enterprise software is "forms first." It assumes the world is clean, structured, and legible. Reality is none of those things.

Operational workflows are conversation-heavy because they have to be.² The inputs arrive as phone calls, texts, photos, and half-complete notes from a tired human at the end of a long day. For decades, our solution was: force humans to translate that into fields so the system can proceed.

That translation is the tax.

The forward-looking bet is that AI becomes the translation layer instead.

In other words, the workflow becomes "conversation first." Humans describe what happened like humans. The system turns it into records, routing, schedules, follow-ups, and a paper trail, without requiring the human to become a data-entry peripheral.

This is not "AI as chatbot." This is AI as the part of the stack that eats the bureaucracy.

Why this matters most in "boring" operational verticals

In consumer contexts, the difference between a good experience and a bad experience is often UX polish.

In operational contexts, the difference is whether the workflow works when it's messy.³

Property operations, home services, construction-adjacent workflows: these markets are full of processes that exist only because software never fully matched reality. That means they have a lot of "paperwork surface area." And that's where AI has leverage.

Not because AI is smarter than people. Because AI is good at:

• extracting structure from messy input
• summarizing and routing
• keeping records coherent across tools
• producing the documentation a system demands

If you apply AI there, you don't just speed up a step. You can remove entire categories of work that exist purely to keep systems in sync.

That is why we think this shift is structural, not incremental.

The mistake to avoid: using AI to spin the same hamster wheel faster

There's a predictable failure mode in every automation wave:

You automate a chunk of admin work, celebrate a capacity gain, and immediately refill that capacity with more volume. Utilization stays pinned. The team stays reactive. The organization stays brittle, just slightly faster at being brittle.

If AI is going to matter, it has to buy headroom.⁴

Headroom is what allows prevention instead of reaction, mentorship instead of triage, and relationships instead of templated exchanges.⁴ It's what lets an organization get better over time instead of just busier.

A diagnostic I use for AI-readiness

If you want to know whether you're pointed at humans or pointed at computers, ask:

When something goes wrong, where does the truth live?

If the truth lives in a conversation, you're human-centric. If the truth lives in a dashboard, you're tool-centric. If the truth lives in a dashboard and everyone knows the dashboard is wrong, you're paying humans to maintain a shared illusion.

AI's best role is to eliminate that third category by keeping the system aligned to reality without requiring constant human patching.

The thesis, stated cleanly

At SCI, we're early. We're not pretending we have a decade of portfolio hindsight. But we do have a point of view about where this goes:

The companies that win in operational AI won't be the ones with the coolest model. They'll be the ones who redesign workflows so the system absorbs the bureaucracy and the humans do the work that actually changes outcomes: judgment, coordination, trust-building, and mentorship.

If you're building in these markets, steal this principle:

Minimize human–computer interaction. Maximize human–human interaction.

That's the correction. And it's overdue.

— jason