# 03. The Governance Layer

This layer is used to ensure we have an appropriate levels of checks that AI is being used responsibly.

## Defining the Utility Ceiling (L\_t)

The **Utility Ceiling (L\_t)** represents the highest *safe* and *technically possible* level of autonomy for a specific tool. The CAI methodology embraces **Right-Leveling**—not every situation calls for Level 4.

Constraints:

1. **Risk Limits (Governance & Compliance):** Requirements that mandate a human-in-the-loop.
2. **Technical Constraints (Infrastructure):** Software architecture and integration capabilities.

### Examples of Utility Ceilings

* **Level 1 Ceiling (L\_t = 1):** Physical/Ad-Hoc Constraints (e.g., Oil rig issue logger requiring onsite physical presence).
* **Level 2 Ceiling (L\_t = 2):** Compliance/Verification (e.g., Medical diagnostics requiring professional sign-off).
* **Level 3 Ceiling (L\_t = 3):** Independent Digital Execution (e.g., Internal bug reporting assigning tickets automatically).
* **Level 4 Ceiling (L\_t = 4):** Systemic Orchestration (e.g., Supply chain AI triggering external vendor orders).

## Risk Limits & Compliance

If a process legally or operationally requires a human to take responsibility for the final output, the system cannot exceed Level 2 (L\_t = 2).

**Low-Risk / High-Autonomy (L\_t = 3 or 4):** Reserved for domains where errors are easily reversible, low-cost, or strictly internal (e.g., micro-transactions, internal IT routing).

## Technical Constraints

You cannot build an Actionable Harness without a place to "hitch" the AI.

* **No APIs / Closed Ecosystems (L\_t = 1 or 2):** Ceiling is firmly capped at Level 2.
* **API/MCP Enabled (L\_t = 3):** AI can receive structured data and return actions.
* **Multi-Agent Frameworks (L\_t = 4):** Infrastructure supports cross-environment agent orchestration.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://cai-framework.gitbook.io/cai-framework/03-governance-layer.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.