Projects are the primary unit of organization inside Melodi. Each project normally represents one concrete AI agent—for example:
  • A customer-facing support copilot
  • An internal sales-assistant bot
  • A staging / QA clone of your production assistant
Because of this, it’s common for the same company to maintain multiple projects, one per product and/or environment (e.g. support-prod, support-qa).

​
Creating a project

  1. Open the Project overview screen in the Melodi web application.
  2. Click Add project.
  3. Provide a name
That’s it—your new project is ready to start receiving data.

​
How projects affect analytics

All dashboards, charts, and metrics inside Melodi are automatically scoped to the currently-selected project. This lets you drill-down into the behaviour of a single agent or environment without noise from the rest of your organisation. However, end-user records are cross-project:
  • If the same user talks to multiple agents (or the same agent across environments) their entire history is visible from a single user page.
  • From a monitoring dashboard point-of-view you will still be looking at one project at a time.

​
Typical workflow

  1. Create a dedicated project for every AI agent you ship.
  2. Optionally create a second project for QA / staging traffic.
  3. Switch the active project from the header whenever you need to inspect a different agent.