This API endpoint will run one of your predefined conversational flows based on the input text’s similarity score to known intents. Calling this endpoint directly is intended for testing purposes - it has already been unified with the /handle endpoint, which uses the large language model to decide whether to invoke a predefined flow or have an AI agent respond directly. The /handle endpoint is recommended for production usage, while this /run_workflow endpoint can be used to test that your conversational flows are triggering as expected before deploying.