Using Impact Assistant
Impact Assistant enables you to ask questions about your data in natural language and receive instant visualizations and answers—no technical expertise required.
Accessing Impact Assistant
Click the Impact Assistant icon inside Spider Impact:
Or inside any stand-alone App you've built:
Asking Questions About Your Data
While dashboards and reports excel at displaying predetermined information, Impact Assistant shines when you need specific answers that aren't already configured. Simply ask questions like "What was the total revenue by product category last quarter?" or "Show me the trend of customer complaints over the past year."
In this example, asking how many devices were sold last year returns a clear answer:
Clicking on the "8,231" number drills down into the Datasets section for further ad-hoc analytics:
Conversational Context
Impact Assistant maintains conversation history for contextual follow-up questions. When you say "break that down by salesperson," Impact Assistant understands you're still discussing last year's sales:
Saving Insights
When Impact Assistant creates a useful chart, use the "Copy to Organization" button to save it permanently. This allows you to reuse assistant-generated charts in dashboards and share insights with your team:
Discovering Data and Insights
Impact Assistant helps you understand your data landscape and discover meaningful insights. You can ask general questions like "What kinds of data do I have?" to see a list of datasets and what they track:
One of Impact Intelligence's most powerful features is proactively suggesting insights to explore based on your data:
Clicking on suggestions like "total sales last month" provides immediate answers:
Getting Software Help
Beyond data analysis, Impact Assistant answers questions about Spider Impact features with detailed guidance and documentation links:
Current Capabilities
Impact Assistant currently:
- Aggregates and filters data from datasets to answer specific questions
- Generates appropriate visualizations (line, bar, and pie charts)
- Suggests potential insights from your data
- Answers questions about Spider Impact features with documentation links
Impact Assistant currently provides answers from datasets only. It cannot access scorecard KPI values or initiative data. Future releases will expand these capabilities.
Managing Conversations
Like all AI assistants, Impact Assistant performs best with fresh conversations. If responses become less relevant or the conversation veers off track, simply close the assistant window and start a new chat rather than trying to correct the current one. This immediately resolves most issues and ensures optimal performance—a quick conversation restart is always more effective than attempting to redirect a confused conversation.
Configuring Impact Intelligence
Enabling Impact Intelligence
Administrators must first enable Impact Intelligence in Application Administration. Toggle "Enable Impact Intelligence" to "Yes" and provide information about your organization in the Organization Context field:
The Organization Context helps the AI understand your data and provide better responses. Include your company name, industry, the types of data you collect, and any big-picture business goals that are universal and don’t change over time. If the AI has trouble differentiating between datasets or finding specific data, additional context here often solves the problem.
Dataset Descriptions
To help Impact Assistant better understand your data context, administrators can add descriptions to datasets. These descriptions help Impact Assistant provide more relevant insights and appropriate visualizations:
When writing dataset descriptions, explain the dataset as you would to a colleague. Include:
- What each record represents
- Types of information tracked for each record
- Why you're tracking this data
- Any important relationships to other datasets
Example description:
Tracks mobile device sales with one dataset record per sale. Most customers purchase a new mobile device once every few years. This dataset has basic info about the sale like its price and date, and it's linked to the customers dataset that has information about who bought it, like the address and the point of contact info.
Good descriptions solve common confusion. For example, mentioning that a dataset tracks both "incident date" and "report date" helps the AI distinguish between them. Similarly, clarifying that "SARC Location" refers to where something happened helps with location-based queries.
API Keys for Self-Hosted Customers
Spider-hosted customers don't need to manage API keys—Spider Strategies handles all billing and infrastructure.
Self-hosted customers need to provide Amazon Web Services (AWS) credentials:
- Create an AWS account and enable Amazon Bedrock
- Generate access keys for Bedrock
- Enter these keys when enabling Impact Intelligence
Alternatively, configure API keys at the server level by adding these JVM parameters to Tomcat:
-Dspring.ai.model.embedding=bedrock-titan
-Dspring.ai.bedrock.titan.embedding.enabled=true
-Dbedrock.access-key=YOUR_ACCESS_KEY_HERE
-Dbedrock.secret-key=YOUR_SECRET_KEY_HEREWith server-level configuration, the API key fields won't appear in the interface—you'll just toggle Impact Intelligence on or off for each instance.