How to get started with the Qlik Answers connector in Qlik Application Automation

[Emile_Koslowski]

Qlik Answers is Qlik's personalized AI assistant that generates on-point answers from your own trusted business content. See this article to learn how to get started with Qlik Answers

This connector unlocks new powerful use cases like indexing new data ad-hoc and taking your action oriented automations to the next level by including unstructured insights. This article explains how the Qlik Answers connector in Qlik Application Automation can be used.

Article contents

• Authentication
• Use cases
• 1. Re-indexing a knowledge base
• 2. Ask questions
• Notes
• Related Content

Authentication

This connector does not require additional configuration to authenticate, it will automatically connect to the automation owner's Qlik account. Whenever blocks of this connector are executed, they will use that account.

Use cases

For this connector's first release, we will introduce blocks that can re-index knowledge bases and ask questions to an assistant. We will deliver more blocks in future releases, do not hesitate to request new blocks through Ideation.

1. Re-indexing a knowledge base

This use case focuses on ad-hoc updates to the knowledge base. The creation and initial configuration of a knowledge base should happen through the Qlik Answers UI as described here:

Once your knowledge base is configured and you've set up a data connection to a cloud storage location, you can then use Qlik Application Automation to add specific files to the Cloud Storage location and trigger a re-indexing. This could be useful in scenarios when you only want to index specific records that might depend on a context.

For this use case, you can use the blocks Add Data Source To A Knowledge Base and Sync Knowledge Base Data Source. The below example will fetch incidents from ServiceNow, store them in a Dropbox folder and then add the Dropbox folder as a data source to a Knowledge Base in Qlik Answers. After this, the data source will be synced to the knowledge base index.

2. Ask questions

By using the blocks Create Thread and Execute Synchronous Prompt, you can ask questions from an assistant to enrich other messages that are sent through automations with unstructured insights. In the below example, these blocks are added to the existing "Notify your team on Microsoft Teams based on a measure" template that can be found in the template picker:

Notes

Don't ask a question immediately after indexing. Asking a question should be done after the knowledge base has completed indexing successfully.

Related Content

• Introducing Qlik Answers
• Qlik.com product page for Qlik Answers

Environment

• Qlik Cloud Analytics
• Qlik Application Automation
• Qlik Answers

The information in this article is provided as-is and will be used at your discretion. Depending on the tool(s) used, customization(s), and/or other factors, ongoing support on the solution below may not be provided by Qlik Support.

[Peter-G-Swart]

Hi

I have a Knowledge Base that contains some (dummy) CVs, with an Assistant on top of the knowledge base.

I have created an automation that calls Qlik Answers Assistant.  It contains the "Create Thread" block and the "Execute Synchronous Prompt" block.

When using the Assistant directly, and using the Assistant via the automation, I get similar, but different answers for the identical (cut&paste) question.  

QUESTIONS: 

Does Qlik Answers receive the identical question when it comes from the Assisant, and when it comes from the "Execute Synchronous Prompt" block?

Why are the answers different?

[davidkoo]

Hi Peter,

There could be a few things happening here.

1. If you are asking the questions in an existing thread from the Assistant, but are asking in a brand new thread in Automations, you may see a different response due to a difference in context.

2. If there's any context or prompt being prepended to the query in Automations, this would cause a difference in response.

3. The LLM itself is nondeterministic, and you may see some differences in response even with the same question.

One thing you should check is if the text chunks are staying the same between the two channels. You can check this through the Assistant UI by referencing the different threads.