Advanced Insights Overview

Overview

Perform natural-language analysis across datasets, surface trends, and take actions on connected systems with Advanced Insights—the conversational AI tool within Expertise Insights. With access to your Kantata data, Expertise Insights dashboards, and external data sources, Advanced Insights allows you to ask questions, explore and analyze data, and discover insights through natural language.

To access Advanced Insights:

1. In the left navigation, hover over Insights, then select Advanced Editor. The Advanced Editor opens in a new tab.
2. Select New chat from the left menu.
   - or - Select Chat agents from the left menu. For an agent in the list, select the Chat link.

 Advanced Insights vs Ask Insights vs Expertise Agent

Several AI features are available in Kantata. The table below provides guidance on when to use which feature.

Using Advanced Insights

On the welcome screen, suggested prompts are listed below the text input. You can select the Refresh icon to refresh the suggestions.

Before entering a prompt, you can configure the following options:

• Agent picker—Choose the agent you’d like to chat with.
• Knowledge mode picker—Select the data you want to chat with.
  • General knowledge—Uses only LLM knowledge. Does not use any knowledge from data resources (i.e. dashboards, datasets, etc).
  • All data—Uses all data and knowledge sources (including dashboards, datasets, topics, Spaces, knowledge bases, connectors, and LLM knowledge).
  • Specific knowledge—You can select a specific asset such as a dashboard, topic, or dataset.
• Select the Add icon to access the following options:
  • File upload
  • Web search
  • Connectors
  • Flow
  • Create—Create documents or visuals. For more information, see the external article.
• Mode picker—You can select Fast, Smart, or Advanced mode.

The following options are available for each agent response:

• Explanation—See detailed information about how the agent arrived at the response, including data sources, assumptions, filters, calculations, and SQL queries. This option helps you validate the agent’s responses.
• Source—Select this option to see the data and knowledge of the response. Source information may also be embedded with in the response text.
• Copy 
• Like—Share positive feedback.
• Provide feedback—Share constructive feedback. A Share your feedback modal opens when this option is selected, where you can add details.

The following options are available at the top of the agent user interface:

• Expand/Collapse
• Chat privately—In private mode, your conversation will not update the agent’s memory or be saved to conversation history. Additionally, the agent’s responses will not be personalized.
• Conversation history—View, access, and continue previous conversations.

How to Create a Custom Agent in Expertise Insights

For most organizations, a single Expertise Insights agent is sufficient to support strategic data analysis for various departments and teams. It is only recommended to create additional agents if you have different teams that need the same data interpreted in different ways. For example, the finance and operations teams need a metric calculated in different ways.

1. In the left navigation, hover over Insights, then select Advanced Editor. The Advanced Editor opens in a new tab.
2. Select Chat agents from the left menu. 
3. In the Create a chat agent section, select the Blank option.
4. On the next page, add instructions in the text field. You can specify characteristics such as identity, tone, personality, and response style.
5. Select Generate. After the process completes, the edit chat agent page opens.
6. Edit the agent’s name using the field at the top.
7. Edit the agent’s description using the field at the top.
8. In the Agent Persona section, you can refine the instructions.
   1. In the Reference documents subsection, you can upload documents to guide the agent’s interactions. For example, skill files, response templates, workflows, methodologies and examples.
9. In the Knowledge Sources section, you can connect the agent to a knowledge base created in Expertise Insights (known as a Space) so that the agent’s answers draw on your own content. For more information on Spaces, see the external article.
10. In the Actions section, you can connect to external tools so the agent can act in your connected apps and take actions such as drafting emails, creating records, and more. For more information on connectors, see the external article.
11. In the Customization section, you can configure a welcome message and suggested prompts to personalize the agent and guide users.
12. As you make changes, select the Update preview button to refresh the agent panel to the right. You can preview and test the agent as you build and refine it.
13. Select Launch chat agent to create the agent.
14. To allow other users to use the agent:
    1. Select the Share button. 
    2. In the modal that opens, search for users, then select a user.
    3. Select a permission level.
    4. Select Share.

Example Prompts

Here are some examples of questions you can ask Advanced Insights to provide inspiration.

• What are the top trends across my delivery KPIs this quarter, and what's driving them?
• Compare utilization in my Utilization dashboard to our target and flag teams below threshold.
• My margin metric dropped last month. Use my dashboards and connected data to explain why.
• Show how our [custom KPI] has trended over six months and correlate it with [external data source].
• Draft an email to my team summarizing this week's delivery dashboard highlights.
• What contributed towards the dip in average project health rank at the end of 2023? What projects, project leads, etc.