This takes about ten minutes, most of it the agent’s analysis.
1

Create your account

Open light.aius.co/login, choose Register, and enter your email and a password of at least 12 characters. Create account signs you in.
The AIUS sign-in card with the Register tab selected, an email address, a password and the Create account button
You land on the Projects page of your personal organisation, named after your email address, for example Personal (ana.lopez). It is yours from the start: create projects in it, publish to it, and invite colleagues later if you want to.
The Projects page of a new account, showing the organisation Personal (ana.lopez), a New project button and the Terminal access panel
2

Install the terminal

On macOS (Apple Silicon) or Linux (x64), run:
curl -fsSL https://light.aius.co/install.sh | bash
Then open a new terminal. Installation covers updating, specific versions and the Python packages the data tools use.
3

Sign in from the terminal

Go to a folder that holds your data and start AIUS. The first time, it asks you to sign in:
cd ~/projects/term-deposit-campaign
aius
The AIUS terminal sign-in screen: Sign in to AIUS, AIUS models and AIUS Light, with browser and device code options
Choose browser to approve in your default browser, or device code (Ctrl+D) to approve from any browser. The terminal then shows an address and a code:
The AIUS terminal sign-in dialog showing Connect this terminal, the address https://light.aius.co/device and the code 7A95-SNJM, waiting for approval
Open the address, check that the code matches and choose Allow access.
The Connect a terminal page in the browser, showing the signed-in account, the same code 7A95-SNJM, what AIUS models and AIUS Light may do, and the Allow access button
One approval connects both the AIUS models and your web platform account. AIUS then offers to keep the folder as a project in AIUS Light, so its working files sync there (Folders and projects).
4

Ask a question about your data

Describe the decision you need to make and point the agent at your files. Ask for the evidence you want to see, and say where the result should be published:
A prompt typed into the AIUS terminal asking for a propensity model on data/bank-marketing-campaigns.csv, published to a new project in the AIUS Examples organisation
The agent inspects the data first, writes its code under output/, runs it in your project’s Python environment and checks its own numbers.
5

Approve, publish and read the report

Before anything leaves your computer, the agent asks. Here it asks to attach its analysis code, executed notebook and aggregate results to the report, and keeps the raw contact rows and row-level predictions local:
The agent asks: May I attach the analysis code, executed notebook and aggregate results to the organisation report? Options: Approve evidence attachments, Report text only, Type your own answer
It then reviews the report, publishes it and gives you the link:
The agent's five-point summary of the result for the marketing team, ending with the link to the published report
Open the link to read the report, its evidence and its attached files on the web platform. Worked example walks through this report section by section.