Demo
From pipeline to AI agent, in fourteen minutes
Seven steps, on open electricity consumption and temperature data: from raw data to a job that an AI agent creates, runs and schedules under your rules.
The scenario
Seven steps
The same journey as a real project, fast-forwarded.
01 · 1 min
The platform, on your premises
graal installed on a Kubernetes cluster, with no administration rights: what your platform team will see, object by object.
02 · 1 min
Sign in through your SSO
Single sign-on on your directory, through a login screen in your organization’s colors.
03 · 3 min
Draw, then export
A pipeline drawn in the low-code editor brings consumption and temperatures together; the export produces readable PySpark code that belongs to you.
04 · 2 min
Go to production
The job runs and its logs stream live; a workflow then schedules it every morning.
05 · 2 min
Find and query
The table appears in the catalog, answers SQL through Trino, then opens in a Jupyter notebook.
06 · 2 min
Measure the model
A regression of consumption on temperature: two training runs compared with the MLflow client.
07 · 3 min
Delegate, then govern
Claude, through MCP, creates, runs and schedules a job under a service account; every call shows up live. Then the project’s roles, audit and costs.
The data
Open, public and verifiable data
Regional electricity consumption (éCO2mix, consolidated and final) and daily regional temperature, published under the Licence Ouverte v2.0. Source: RTE and Weathernews France via ODRÉ.
Contact
Book a demo, get a pack or an estimate
This form prepares an email in your own mail app: nothing is sent until you send it yourself, and nothing goes through this website.
Prefer to write directly? contact@graal.systems. What happens to your data: privacy policy.
Would you rather start with your own data?
The 8-week pilot takes one use case end to end, in your infrastructure.