For the last 30 years we humans have been really busy managing and working in the software we bought.
We even created jobs dedicated to the management of that software: 'salesforce admin', 'marketing opps', 'sales opps', etc.
Then there were a cadre of jobs that managed humans doing work inside the software: BDR, support rep, CSM, marketing analyst, sales rep, etc.
What’s changed now is that AI systems don't really need humans help to run, and don't even need humans do do that work. As this is happening there are a lot of roles, tools, and assumptions that are starting to feel less stable.
You can see it first in entry level roles.
Entry-Level Work Is Disappearing
Gen Z isn’t “bad at AI.” They’re the first generation being replaced by it.
That’s the thing hiding in the LinkedIn data if you don’t dress it up. Boomers and Gen X see AI as leverage. They already have jobs, credibility, and income. AI just makes them faster and work less.
Gen Z sees AI show up right as they’re trying to enter the workforce and the first thing it does is eat entry-level work: SDRs, Support Reps, Analysts, Assistants. Junior creatives, and Ops roles. These are the jobs you used to use to get started.
So no, they’re not “less confident.” They see that they are being squeezed by AI.
Most companies rolled out AI without thinking through what they were actually introducing. No clarity on what humans are supposed to own once the machine shows up. No path. No handoff.
Just a lot of “figure it out” energy from leaders who already crossed the bridge.
Of course Gen Z is skeptical. They're being handed powerful tools with no job training or mentorship like we had for the last 50 years. Then they are told to feel grateful for the job.
If you want confidence, stop telling them to “learn AI.”
Show them:
- which jobs still belong to humans
- what kind of judgment is needed
- how using AI makes them more valuable, not easier to cut
AI doesn’t scare Gen Z, but being disposable does.
If you’re a leader, you need to fix the system. And if you’re early in your career, don’t wait for permission. Learn the tools anyway. The jobs disappearing first weren’t strategic.
When that work stops needing a person, it doesn’t just affect careers. It changes what software is actually used for.
The Day Humans Became Observers
Moltbook is being described as “Reddit for AI agents.” ... but its WAY bigger that that.
This is not a new AI social network. To me it feels like a stress test. A test for a belief we’ve all been lying to ourselves about: that humans are still the center of digital systems.
We are not.
If you’ve actually spent time on Moltbook, you notice something fast. Humans don’t post. We can't comment or moderate. We just observe.
Most AI stories are still human-centric:
- AI helps me write
- AI helps me research
- AI helps me code faster
Moltbook skips that phase entirely. The openclaw agents aren’t helping humans do things. They’re coordinating with each other to do things that they "want" to do.
For real, pause reading this and go check it out.
Different things now start to matter.
Agents can now communicate, evaluate, and reinforce each other without human permission. Governance should now matters more than models or really anything else.
This shows what happens when coordination detaches from people. It’s the first glimpse of systems where humans are no longer participants. Just observers. If you’re building AI and still assuming humans sit at the center forever, Moltbook is your warning shot.
That sounds abstract until you see the same pattern show up in much simpler, real-world forms.
AI Is Hiring People to do the Physical Work
AI is now hiring humans to do physical work. Not sure if I should laugh or cry. Possibly both.
Rentahuman.ai lets AI agents post bounties and hire people for physical work. Some of the examples are unintentionally hilarious. Running errands. Pressing buttons. Going somewhere so the AI doesn’t have to.
The tasks are small. The implication is not. Which means:
- The agent scopes the task
- The agent sets the price
- The agent decides when it’s “done”
Pretty wild if you ask me.
We spent decades building software to “support” humans. Now the software is asking us to go pick something up. No anger or drama. Just a task, a price, and a notification. Which is somehow the funniest and most unsettling part of this whole thing.
That moment matters. Because once software starts assigning work instead of supporting it, execution moves into the system. And when execution moves, the economics reprice.
That’s exactly what the markets reacted to next.
Markets Are Repricing Software
Claude Cowork launched and wiped out $300B in market cap across SaaS. Now it’s getting called a “SaaSpocalypse.”
Some people think it’s an overreaction. I don’t.
The selloff happened because investors finally saw something clearly: software is losing control of execution.
Claude Cowork didn’t beat SaaS on price or features. It removed the need for a human in the middle. That’s why Salesforce, ServiceNow, Adobe, Workday all bled at the same time. Markets don’t do that unless a core belief snaps.
The snapped belief: “Recurring revenue = predictable value.” This is wrong now. When intelligence executes work across systems, licenses stop mattering. Outcomes do.
As Chamath Palihapitiya put it, the SaaS meltdown has started. He’s right, but it’s not a collapse. It’s a reallocation. Goldman Sachs already modeled where this goes: By 2030, most software economics flow through agents, not seats.
Software becomes the container and intelligence becomes the product. Seats don’t disappear overnight. They just stop being the thing anyone cares about.
Markets noticed this shift first because markets watch cash flow, not culture. But the people who actually feel it next aren’t investors. They’re operators.
Execution Moves to the Operator
Most enterprise software is a tax. A daily inconvenience tax paid by the individual doing all of the work so the org could standardize, track, and report.
This week's panel at Silicon Slopes made something painfully clear: AI changes who software actually serves. The last 20 years, software rewarded management.
This next decade, software rewards the operator. We all need to become that operator.
When intelligence can actually execute across systems, a few things break fast:
- Seats matter less. Results matter more
- Teams stop growing in neat ratios. They get smaller
- A lot of the glue layers we’ve all learned to live with just… aren’t needed.
What surprised me wasn’t the tech. It was where the power ends up. The person who understands the workflow and knows how to use agents don't need permission anymore. They don’t wait on handoffs. They don’t need a bigger team. They just ship.
One comment from the discussion stuck with me:
“The only way to be a founder now is to go back to being an individual contributor.”
You can already see it playing out. Founders who stay close to the work move faster. The rest spend their time explaining delays.
You can’t run a company like it’s 2019, with managers managing and teams stalled between steps. Smaller teams are moving faster. Not because they’re smarter. Because they’re closer to the tools and less attached to how work used to flow.
Stop adding software to feel organized.
Start building leverage where the work actually happens.
Get in the tools or someone else will.
Once execution lives with the operator, software won't just coordinate work anymore, it will do it.
Software That Does the Work
SaaS is dying. Not because it failed, but because never actually helped us.
For 20 years, SaaS companies sold software and called it service. But it wasn’t a service, It was just a license fee, a login screen, and a human doing all the execution. That era is over.
AI doesn’t just support work, it does the work.
- Instead of sequences, it generates pipeline.
- Instead of helpdesks, it resolved tickets.
- Instead of sourcing tools, it sources new hires.
This isn’t SaaS 2.0.
It’s the end of renting tools and the beginning of hiring AI teammates to deliver outcomes. Call it Outcomes-as-a-Service.
Growth stops being about managing software and hiring more people, and starts being about getting more work done.
Headcount stops being how companies scale and software stops being the value. Organizations are now getting rebuilt around generating real output.... This is the Autonomous Organization. And that’s why we’re building Atonom.
With Atonom, you hire Cloud Employees (advanced AI agents that learn your business) that integrate with your systems and deliver real results. They aren’t tools. They’re AI teammates.
Cloud Employees are the infrastructure of the Autonomous Organization. This isn’t a trend. It’s the direction work is moving. You can watch it happen. Or you can build it with us.
Learn more about Atonom here.
Zoom Out
The last gen of software helped organizations manage, hire, and track people. The next gen is about producing results.
Work won't entirely disappear. But the value stops coming from supervising process, and will start coming from understanding systems well enough to let them run without interference.
You can already see which teams get it. They’re smaller, more nimble and with less organization depth. Winning organizations (the autonomous enterprises) have leaders that are MUCH closer to the tools and are fluent in AI.
