How Frontal fits together

An intelligent system is more than a model call. It needs trusted context, an explicit objective, bounded tools, durable execution, and evidence that the result was appropriate.

CopySources → Data → Ontology → Context → Agents and workflows → Runs
                                           ↓
                         Evaluation, observability, security, and governance

Frontal keeps these layers connected. Data and ontology establish what a record means; storage preserves the artifacts and retrieval state around it; agents and workflows use that context to perform work; runtime controls execute that work; evaluation and observability show whether it was correct and safe.

The layers are useful independently, but the value is in their shared boundaries. An agent, a workflow, and a person should resolve the same customer and be subject to the same permission decision.