In this article
The November 2025 Forsta Visualizations Release introduces powerful enhancements focused on AI-driven insights, usability improvements, and expanded data integration.
Key updates center on the Research Agent, Data management, Data connectivity and a new Image Map import and visualization workflow.
StoryTeller Enhancements
Research Agent
The Research Agent now provides a more interactive, intelligent, and contextual AI experience through the following updates:
Dynamic Prompt Generation
Introduces automated, context-aware prompt creation based on the active slide’s content.
When “Dynamic Prompt Generation” is selected, the AI analyzes the slide and proposes several relevant prompt suggestions.
Each generated suggestion functions as a prompt request, processed through AI to deliver tailored insights.
Users can send a suggested prompt directly or edit it for better precision before submitting.
Continued Conversations
Users can now carry on a conversation thread within the same AI session.
Enables iterative exploration—ask follow-up questions or refine prior insights based on earlier AI responses.
Despite this conversational flow, the AI remains stateless: each request includes the slide image, current, and previous prompts and responses, ensuring context continuity without persistent memory.
Multi-Slide Analysis
The Research Agent can now analyze multiple slides simultaneously.
Multi-slide selection is available via a dropdown under the screenshot preview.
Works in both horizontal and vertical report views, maintaining consistent logic and user experience.
Improved Screenshot Logic
Enhanced screenshot handling supports multi-slide and vertical layouts, improving flexibility for complex reports and multi-dimensional analyses.
Base Size Limits for Word Clouds and AI Summaries
A new safeguard improves data reliability and interpretability:
Introduces automatic hiding of data when the base size falls below a defined threshold—mirroring Respondent Table behavior.
When data doesn’t meet the threshold, the object displays a localized information message, replacing the visualization.
The text message is defined in each object’s settings and can be translated across all supported languages.
The design of this alert mirrors the “NoData” message style for visual consistency.
Data Management
AI Computes – Preview of Prompts
This feature allows users to test and validate AI prompts before full execution, minimizing errors and fine-tuning analytical performance.
The Preview of Prompt option enables testing a prompt on a subset of respondents.
The preview sends selected data and the written prompt to AI, returning reference data that helps verify intended results.
Users can choose how many respondents to test via a dropdown selector.
Preview output appears directly below the prompt field for immediate review.
Data Connectivity
Import Image Map Data from Forsta Decipher
A major addition for integrating visual, spatial response data:
The Image Map question type from Forsta Decipher lets respondents answer by placing markers on an image (e.g., selecting liked or disliked areas).
How It Works in Visualizations
When imported, Image Map questions are treated as open-ended questions.
Each marker (e.g., Like, Neutral, Dislike) is imported as a separate question.
The image itself is imported separately and displayed automatically when the corresponding Image Map question is selected.
New Object Type: Image Map
A new visualization object—Image Map—is now available in StoryTeller.
It only displays questions linked to Image Map variables from Forsta Decipher.
The setup mirrors other objects: choose the question, then configure Filters, Settings, and Layout tabs.
Customization Options (Settings Tab)
Adjust heatmap colors to visualize response intensity.
Add base size info and a color legend.
Includes a toggle to show or hide the color overlay over the base image.
Display Notes
Turning off the color map automatically hides the legend.
The object size depends on the original image dimensions.
Base info, color legend, and toggles appear within the frame but outside the image itself.
For best results, use images no larger than 1280×720 px (PowerPoint slide size) to maintain layout quality.