Most AI and ML work begins with open-ended experimentation, and Anaconda Platform supports this type of work out of the box. You can spin up a workstation, run flows, and scale compute as needed.
This experimentation work still benefits from all the core functionality of Metaflow, including versioning, artifact tracking, and namespacing, without doing anything extra. However, as the work matures, it needs structure—a shared home for the code, data, and models that make up the system, repeatability you can hand to a teammate, and a delivery pipeline for getting changes into production safely.
That is what a project adds. Projects bundle your flows, deployments, and assets under Git-backed branches, with CI/CD integration built in, similar to how modern web development platforms like Vercel or Render work, but purpose-built for AI and ML.
The default project
Work that does not belong to a project lives in the default project, a space for quick experiments and exploratory work. Use the context picker in the header to switch between the default project and any project you create. The default project shows all runs, workflows, and deployments, regardless of whether they belong to a project.
Features specific to projects, such as the project overview page, evaluations, and assets, are not available in the default project.
When to set up a project
Projects are quick to create and inexpensive to keep, so you don’t need to set a high bar for making one. For instructions, see Setting up a new project. 
