Witty Governance Agent Engine

A policy with Witty governance agent engine, allows users to provide natural language text which will be used by an agent to evaluate resources.

This allows validations to be defined using an agent, which is nothing more than a Large Language Model (LLM) that receives the resource's descriptor and the user's natural language text used for evaluation. If the descriptor satisfies what the user expressed in natural language, the agent gives a positive result, otherwise it highlights errors not allowing the policy to be satisfied.

To work best, policies must have concise text. Guidelines on how to write policies in the most efficient way will be provided soon.

The following are examples of evaluation results of a Witty policy. This policy requires that the data product domain equals “finance” to be satisfied.

The data product domain is set to "finance", satisfying this policy.

The data product domain is set to "marketing", not satisfying this policy.

The Open API specification used by the Witty Governance Agent Engine is available at Witboost Extensions Points API.