# Workflows Concepts (private)

Generated on 26 Mar 2026

Workflows automate machine learning tasks, combining GPU instances with an expressive syntax to generate production-ready machine learning pipelines with a few lines of code.

[Environment Variables](https://docs.digitalocean.com/products/paperspace/workflows/concepts/environment-variables/index.html.md): Environment variables are one way to pass information to and from Workflows, and between the jobs within them.

[Understanding Inputs & Outputs](https://docs.digitalocean.com/products/paperspace/workflows/concepts/understanding-io/index.html.md): The steps in a Gradient Workflow can communicate with other steps using inputs and outputs, which can be datasets, volumes, or strings.

[Using YAML for Data Science](https://docs.digitalocean.com/products/paperspace/workflows/concepts/use-yaml-for-data-science/index.html.md): Examples of YAML syntax used in Gradient Workflows.