
Mosaic AI Agent Framework
The six pillars
Each bar is the score; the bracket is its 95% range.
Can you see what it costs?
Can you see how much it's used?
Can you tie cost to one agent?
Can an agent on it report out?
Can you control spend and access?
Does it work on small plans?
Where the evidence comes from
Once three independent fleets report on a pillar, desk research counts for at most a third of it. FleetRegistry's own fleets never count.
Controls on record
Budgets alert by default; 'Block usage' exists only for Unity Gateway (LLM/agent requests) and is enforced approximately, and Databricks says not to rely on it as an absolute cap; provisioned throughput and compute budgets are alert-only. Token permissions, auto-scoped/applied token scopes and the Token Management API (list/delete) give scoped revocable identities; audit via system.access.audit (Public Preview); SCIM requires Premium or above.
History
Run agents on Mosaic AI Agent Framework?
Give each one a permanent number, a named person who answers for it, and a record of what it can touch and what it costs. Free.