Most teams start with a single shared API key for AI access. It works — until someone leaves, costs spike with no way to trace them, or you need to give one role access to a better model. Per-resource AI assignment in Maverick solves all of these problems before they become incidents.
1. No shared credentials means no shared risk
When every employee uses a shared API key, one compromised credential exposes everyone's AI access. Assigning individual API keys means each person authenticates independently. If a key is compromised, you revoke one key — not the key that powers your entire organization's AI workflows. Per-resource credentials are the AI equivalent of not sharing passwords.
2. Transparent, per-person billing
AI usage billed to a shared key shows up as one undifferentiated line item in your provider account. Per-resource keys let you see exactly how many tokens each person or team is consuming, which roles generate the highest AI costs, and whether the investment is paying off in productivity for the people who use it most. This data is essential for making informed decisions about model upgrades and AI access policies.
3. Match model capability to role requirements
A project executive analyzing multi-project risk may benefit from a frontier model with strong reasoning and long context. A team member doing routine task scheduling works fine with a faster, cheaper model. Per-resource model assignment lets you right-size AI capability to each role — paying more only where it generates proportional value. One size rarely fits all when it comes to AI models.
4. Easy offboarding with no side effects
When an employee leaves, deactivate their resource record in Maverick and their AI key stops being used. No other users are affected, and the key can be revoked at the provider with no coordination required. Compare this to a shared key scenario, where offboarding requires rotating the credential for every remaining user — a process that disrupts anyone who hasn't updated their settings.
5. Workgroup-level defaults with individual overrides
Maverick lets you set an AI provider at the workgroup level and override it for specific individuals. Configure an entire department in one operation, then make targeted exceptions — a different model for the team lead, a higher token limit for the data analyst — without reconfiguring everyone else. This layered approach makes initial setup fast and ongoing management practical at any team size.