AI in 2026
It started with chatting. Then we gave the agents tasks, then other services to work with. Developers went from a code editor to the terminal. Now two new doors are open, the agent writes the thing for you, or manages other agents. More authority to the agent, less technical you have to be.
- Size
- A quick one
- Short link
- piy.show/s119
Is AI getting harder?
Every few months a door opens that was closed before. The question is how you keep up, so let us review where we are today.
I drew it as a ladder on screen: the chatter, the subscriber, the developer, the architect, the CEO. Each step gives the agent more authority, and each step asks less technical skill of you. That is the surprise, and the whole point of this Short.
Which step of the AI ladder am I on? Prompt it yourself
Where am I on the ladder from chatting with AI to running agents? Ask me what I do with it today, then tell me the next step and what it needs.
Chat, then tasks, then other services
It all started with chatting with agents. A question in, an answer out, on your phone. Then we gave them different tasks to do. Nowadays we want to integrate them with other services, so the agent can read the invoice, log in to the portal, and act, not just talk.
Most people are still at the first step and pay for the second: a subscription on a phone or in a browser, with limited access. The agent can tell you what to do. It cannot do it, because it cannot reach anything.
Give it access to one service you use every day. That is the move from chatting to delegating.
What changes when the agent can reach my services? Prompt it yourself
Here is one task I do by hand every week. Do it for me, and tell me what access you need to do the whole thing next time.
The developers' side of the story
The other side of the story is the developers. They are using AI to massively boost their productivity. First inside the code editor, then in the terminal, where the agent works in the whole folder and runs the commands itself.
That is the side I come from, and it is where the real power lives: on your computer, in a terminal, with the agent allowed to act.
You do not have to be a developer to sit there. You have to be willing to open the terminal once. The agent does the typing.
Why is the terminal the stronger seat? Prompt it yourself
Set up the agent on my computer so it can work in a folder and run commands for me. Explain every step before you do it.
I can help you build yours
I build things like this with an AI agent, on my own computer. In a private session I help you do the same with your idea. Or bring it to the show, free.
Two new doors
Recently two new ways have been unlocked. You can ask the agent to write the things for you. Or you can ask the agent to manage other agents, and sit above them like a CEO.
Do you see what is happening? More authority for the agent, and less technical you have to be. The ladder goes up, and the skill it asks for is not code. It is saying what you want and checking what comes back.
What does it mean for an agent to manage agents? Prompt it yourself
Plan this project as a team of agents. Tell me what each one would do, what I would check, and start with the first one.
Start with the next step
Find your step on the ladder. Take the next one, not the last one. Give the agent one real task and one real service, and prompt it yourself.
What do I do next? Prompt it yourself
Here is one real job from my week. Do it, show me, and tell me what the next step up would need.
Prompty