← Back to the start

Workshop

Build one in a workshop

This is for people who want help setting up their first agent and want to understand what they are running. Make the choices together, connect only what the job needs, and get one real task working before adding more.

01

What we do together

You do not need to know what a model key or a virtual machine is before starting. Each choice is explained when it comes up.

  1. 1

    Give the agent a job

    Bring one thing you genuinely want help with. A real task is much easier to test than a vague idea of having an assistant.

  2. 2

    Rent the computer

    We will use a small cloud computer so the agent can keep running when your laptop is closed. You do not need to buy another computer.

  3. 3

    Choose its brain and instructions

    We will choose a model, set its basic instructions, and make sure the job is clear before adding memory or app connections.

  4. 4

    Choose where to talk to it

    That might be the built-in interface or a chat space you already use. We will start with the simplest option that works.

  5. 5

    Run the task

    We will watch what it does, fix the obvious problems, and write down what still needs work. The point is to leave with something useful, not a pretty empty setup.

02

What to bring

One task and a rough idea of what information the agent would need. Do not connect your whole inbox just because you can.

A good first task

"Research this question and keep the sources" is testable. "Run my life" is not. We can make the agent more ambitious after it earns that trust.

Your access line

Decide what is off limits for the first session. If the task works with public information or a file you upload, we can leave your accounts alone.

Workshop status

The next workshop is not scheduled yet

The date, format, and price will be clear before anyone is asked to pay. Share what you want to build so the next session can be shaped around real jobs.

Share what you want to build