Steps to Trigger a Discovery Run

Updated 

How to trigger a Discovery Run:

  1. Open the Conversational AI Persona App and go to "AI Tools". Click on "Discovery Runs"

  2. To start a new discovery run, click on "Add Discovery Run" in the top right corner and select the desired option.

    1. If you want to use existing data from Sprinklr, select the "Intent" option.

    2. If you prefer to upload data using Excel, choose "Intent via Excel" and download the example template

  3. After applying the relevant filters, click "Start Discovery Run" to begin the process.

Discovery runs may take time to complete. Once they are finished, you can view the results and create intents by following this article.

Note: If a Discovery Run lasts more than one day and is still ongoing, it is considered a failure. In this case, please open a support ticket and start a new Discovery Run.

Advanced Configuration 

Intent Cluster Discovery Configuration: Set the level of granularity for clusters created during Discovery Run. Choose High for more specific clusters with less data per cluster, or Low for fewer clusters with more data per cluster, at the cost of specificity.To create a seed cluster, provide examples of known clusters and expected customer phrases. The AI will create a cluster based on similar phrases.

Discover Workflows

We recently launched Discover Workflow as a beta feature which allows users to explore bot workflows by analyzing conversation history between customers and agents in Sprinklr. This feature can help users uncover new bot flows and enable efficient self-service options.

To access this function, go to "Discovery Runs", click the dropdown icon by "Add Discovery", and then select Workflow.

Conclusion

Sprinklr's Discovery Run tool simplifies the creation of conversational AI solutions by automatically identifying user intent from past conversations. This reduces the time and resources required for brands that are building virtual assistants. By following the Discovery Run process, you can gain valuable insights and enhance conversational models for an even better user experience.