> ## Documentation Index
> Fetch the complete documentation index at: https://irisdocs.prescientlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Client AI Governance

> Create lightweight ownership and review mechanisms so AI initiatives remain controlled after the consultant leaves.

Governance should answer a practical question: **who is responsible for each AI use case before, during and after deployment?**

## Minimum use-case record

For each production AI system record:

* business owner;
* technical owner;
* purpose;
* affected users;
* data sources;
* model/provider;
* autonomy level;
* human oversight;
* material risks;
* approval status;
* performance KPI;
* review date.

## Decision rights

Clarify who can:

* approve a new use case;
* approve data access;
* approve production release;
* change model/provider;
* change system prompts or tools;
* pause the system;
* accept residual risk.

## Review cadence

Review high-impact systems more frequently than low-risk internal assistance. Trigger an additional review after material changes to model, data, workflow, regulation or user population.

## Connect governance to delivery

Do not create a governance framework that nobody uses. Integrate reviews into the project lifecycle:

**Opportunity → risk assessment → design → test → approval → production → monitoring → periodic review**

## Consultant exit condition

Before handover, ensure there is an internal owner, monitoring mechanism, documentation and escalation route. A production AI system with no accountable owner is unfinished work.