Go back to the flight-price agent from Lesson 1. For it to message you the moment a price drops, something has to run that check every morning, whether or not you are at your desk, whether or not your laptop is open. That simple requirement is where a lot of first agents quietly die.
A chat session is not a running agent
When you use a chatbot, it only exists while the tab is open and you are typing. Close the tab and nothing is watching, waiting, or working. That is fine for a conversation. It is useless for an agent whose whole value is doing things while you are busy or asleep.
Why your laptop is the wrong home
The obvious idea is to run the agent on your own computer. It works for a quick experiment, and it is a fine place to learn. As a permanent home it falls apart quickly:
- It stops the moment your laptop sleeps, closes, or loses wifi.
- It cannot wake up at 7am to send your digest if the machine is off.
- You become the operations team: updates, restarts, and crashes are all yours.
- Your credentials and the agent's work sit on a personal device that travels with you.
What an always-on home gives you
- It runs on a schedule, so your 7am digest actually arrives at 7am.
- It reacts to events, like a new email or a price change, the instant they happen.
- It keeps going for long tasks without babysitting.
- It keeps secrets and work off your laptop, in one place you can monitor.
You can set this up yourself by renting a server and configuring it, which is real work and real maintenance. Or you can use a managed service that keeps the agent online for you, so you spend your time on what the agent should do rather than on servers. Qoren is one such service: it runs your agent in a managed, always-on environment so you never touch the plumbing. Either way, the lesson is the same. Decide where your agent lives before you fall in love with what it can do.
Now that your agent has somewhere to live, the next lesson gets practical: we design your first one, on paper, in plain language.