Skip to content
Nappai Documentation

Vectara

Vectara is a component in Nappai that allows you to store and search your documents using a powerful vector search engine. This means you can find documents that are similar to what you’re looking for, even if they don’t contain the exact same words. Think of it as finding documents based on their meaning, not just keywords.

Relationship with Vectara

This component connects directly to the Vectara vector database service. It uses Vectara’s API to store and retrieve your data, allowing you to leverage Vectara’s advanced search capabilities within Nappai.

Inputs

  • Embedding: This input provides the vector representation of your data. Think of it as a numerical code that describes the meaning of your document. You don’t need to worry about creating these; Nappai handles that for you.
  • Ingest Data: This is where you provide the documents or data you want to add to the Vectara database. You can add multiple documents at once.
  • Search Query: This is the text you type to search for similar documents. The more descriptive your query, the better the results.
  • Number of Results: This lets you specify how many search results you want to see (default is 4). This is an advanced setting.
  • Credential: This is your login information for accessing the Vectara service. Nappai will guide you through setting this up securely.

Outputs

The component produces a list of documents that are most similar to your search query. These results are presented as data objects within Nappai, ready for you to use in your workflows. You can then use other Nappai components to process or display these results.

Usage Example

Let’s say you have a collection of customer support tickets. You can use Vectara to store these tickets. Then, when a new ticket arrives, you can use the Vectara component to search for similar past tickets. This helps your support team quickly find solutions and improve response times.

  1. Add Data: Upload your customer support tickets to the Ingest Data input.
  2. Search: Type a summary of the new ticket into the Search Query input.
  3. View Results: Nappai will display a list of similar past tickets, helping your team resolve the issue efficiently.

Templates

[List of templates where the component can be seen and its configuration - This section requires information not provided in the original prompt.]

  • VectorStoreInfo: Provides information about the vector store being used.
  • Self Query Retriever: Uses a vector store (like Vectara) and an AI model to generate search queries.
  • [Other Components Listed in Prompt]: These components can be used before or after Vectara to process data, send notifications, or perform other actions. (Detailed descriptions would require individual component documentation).

Tips and Best Practices

  • Use descriptive search queries for better results.
  • Ensure your data is properly formatted before ingesting it into Vectara.
  • Experiment with the Number of Results setting to find the optimal number of results for your needs.

Security Considerations

  • Always use strong and unique credentials for accessing the Vectara service.
  • Regularly review and update your Vectara credentials as needed. Nappai’s security features will help you manage this securely.