Imagine you could redesign your company knowing agents exist.
What would you change?
Would you still buy Tableau?
Would you still buy a dashboard builder that needs a person sitting in the middle of every report?
Would your tools still be designed mainly for humans clicking through screens, or would they expose interfaces agents can use directly?
That is the shift AI forces.
It makes you look at your company and ask where the real constraint is. A lot of the time, the constraint is the human in the loop.
That does not mean humans should disappear.
Humans are good at judgment. Direction. Taste. Context. Knowing why something should exist.
But humans are often wasted on the how.
Pull this report. Clean this spreadsheet. Check this dashboard. Write this update. Move this information from one tool into another. Turn the same analysis into a deck again.
That is not judgment. That is motion.
The agent should handle more of the motion.
So instead of asking, "How do we add AI to this workflow?" the better question is:
"Should this workflow exist in this form at all?"
That one question changes the conversation.
You stop looking for places to sprinkle AI around the company. You start looking at the company itself.
Why is this process here? Why does this dashboard need a human owner? Why does this internal service have a human interface but no agent interface? Why does this team need another coordinator instead of better leverage?
Once you start looking at workflows this way, the whole company starts to look different.
Do you need the same hierarchy of people managing people?
Or can the company be leaner, with specialized operators using agents to run more of the work?
This is already happening at the edges.
Anthropic's growth marketing function was reportedly one person using AI for a long stretch. That does not mean every growth team should be one person. It means the shape of the team is now up for debate.
The old default was simple:
More work means more people.
The new default is less obvious:
Should you hire a specialist? Should you hire a generalist who can operate across tools? Should you build an agentic workflow instead? Should one sharp operator run a team of agents before you add another full-time person?
This is why "AI transformation" is usually the wrong frame.
Transformation sounds like taking the company you already have and adding AI on top.
A chatbot here. A copilot there. A few automations inside the same old workflows.
That might save time. But it avoids the better question:
If we were building this company today, would we build it this way?
Most companies would not.
They would not buy the same tools. They would not design the same reporting loops. They would not route so much work through humans by default. They would not hire the same way.
AI gives you permission to ask the annoying question again.
What is this company actually trying to do, and what is the smallest, sharpest version of the company that can do it?
That is where the redesign starts.