Azure AI Foundry Project
Does ZopNight manage Azure AI Foundry Project?
AI Foundry projects are the containers that tie model deployments and usage back to an owning team. ZopNight discovers every project under its AIServices account, records provisioning state, and uses the project boundary for team-level attribution of AI spend, so a token bill can be traced to an owning project rather than a shared account.
Rules that fire on Azure AI Foundry Project
No active rule family targets Azure AI Foundry Project today. Rules that used to are retired, and retired rules publish no pages and fire no findings. Scheduling and permissions coverage are unaffected.
At a glance
| Field | Value |
|---|---|
| Scheduling notes | discovery and topology only. |
AI Foundry projects are the working containers where teams build with models, agents, and data. Projects tie deployments and usage back to the owning team for accountability.
Projects as the attribution boundary
Meters bill at the Cognitive Services account level, not the project level, so on a shared account the invoice alone cannot say which team spent what. The project is the natural place to recover that answer. Each project scopes a team’s deployments, agents, and connections, and carries the membership that says who can create billable things inside it. When several teams share one AIServices account, the project boundary is the only accountable line between their workloads.
How ZopNight ties projects to spend
Discovered via the AI enricher under its hub or account, providing team-level attribution of AI deployments and their spend. For each project ZopNight records the ARM identity and provisioning state, then hangs the project’s own connections beneath it. A project-level connection to an external model provider is a different governance fact than an account-level one, and the topology keeps them distinct. Projects are structure rather than compute, so there is nothing to schedule; their value is making the account’s spend explainable.
Project-level hygiene problems
Inventories keep surfacing the same three. Dead projects from finished experiments, still listed as owners of live deployments. Projects whose membership outlived the team, where people who moved on can still deploy models that bill the shared account. And the one-project-per-person anti-pattern, which turns a manageable account into dozens of attribution lines with no clear ownership at all.
Opening projects in the AI Foundry portal
The AI Foundry portal is the working surface: pick the account or hub, and its projects list appears with each project’s deployments, agents, and connected resources. From the Azure portal side, a Foundry account’s page enumerates the same projects as child resources, which is useful when auditing an estate subscription by subscription rather than project by project.