RAG pipelines

RAG pipelines draw from the Public Pattern Library and pipelines in your environment (Org) to generate a pipeline. The Generate a Pipeline workflow enables you build pipelines based on simple text prompts. You can also leverage existing pipelines to generate a new pipeline. When you enter a prompt to generate a pipeline, the response gives you the option to select example pipelines. You can bypass these selections by clicking Continue without example pipelines.

RAG pipeline workflow

The following steps describe the workflow for using RAG pipelines to generate a new pipeline.

1. Open SnapGPT and click Generate a Pipeline.
2. Enter a prompt. For example: Generate a pipeline to send Salesforce data to an s3 bucket
3. When SnapGPT responds, click Select Pipeline to open the Select patterns or pipelines to improve quality dialog.
4. Scroll through the pipeline examples from Project and Pipelines list. SnapGPT offers two sources from which to choose pipeline examples:
   • SnapGPT Suggestions: Most closely related pipelines.
   • Project Space: Pipelines in your environment.
5. For each selection, click the arrow keys to add or remove from the Select Pipelines window. You can choose up to 5 selections.
6. When you finish, click Use selected to proceed with the pipeline generation. The RAG window closes, and SnapGPT generates the pipeline.
7. Three different pipelines are generated for you. Click on any one of them to open in the existing or a new tab.