Skip to main content
Deployment on Prisme.ai runs on two independent cycles. Keep them separate in your head and the rest is simple: the platform moves through its own environments, and each use case has its own dev and prod workspaces on the production environment, versioned by its own Git.
Platform versions flow Dev to Preprod to Prod with go/no-go, run by the Platform Squad with CI/CD. A use case runs on the Production platform environment as a Dev workspace (Git read/write) that pushes to a use-case Git repo, from which a Prod workspace reads (read-only). The use-case lifecycle is decoupled from platform upgrades.

​
Platform environments (self-hosted)

The platform itself moves through isolated environments on your own landing zone (AWS, Azure, GCP) or on-premises. A minimum of two is recommended, three when critical use cases justify it. This cycle is run by the Platform Squad with CI/CD. Each version is promoted Development, then Pre-production, then Production, with a formal go / no-go at each step. Critical use cases must be testable on every environment before a version reaches production. See self-hosting updates.

​
Deploying a use case

A use case does not get its own platform environments. It runs on the production environment, as two workspaces synchronized by a dedicated Git repository. Its lifecycle is decoupled from platform upgrades, and Git guarantees reproducibility.
1

Dev workspace (Git read / write)

Where the team builds and iterates, on the production platform environment. Connected to the use case’s dedicated Git repo in read / write.
2

Controlled promotion

Versions are promoted through the CD pipeline and / or the Builder UI. There is no manual change in production. See deployment.
3

Prod workspace (Git read-only)

Connected to the same Git repo in read-only, so production always runs a validated, versioned release.
Critical use cases during a platform upgrade. When the platform moves to a new version, critical use cases are also validated across the Dev, Pre-production, and Production environments. This validation is carried by the Platform Squad, not by the use-case teams.

​
Workspace or self-care?

Not every use case needs the two-workspace, Git-versioned pipeline. Choose per use case.

Workspaces (Git-versioned)

Dev and prod workspaces synchronized by Git. Required when every change must be traced and versioned.

Self-care (no-code)

Agents created and updated directly in the product interfaces by business teams, with no workspace, repo, or pipeline. Enough for lighter, self-service use cases. See Agent Creator.

​
Next steps

From a use case to production

The full delivery lifecycle in phases

Builder deployment

How promotion and Git sync work in practice