The release of Sprint 93 marks a significant leap in scalability, data governance, and improvements to the editor experience. This update solidifies our bulk management capabilities, refines data governance, and introduces a more robust, host-driven architecture for the Story Builder.
Driving Scalability with Assets Group Actions
As part of our commitment to high-volume efficiency, we have introduced Assets Group Actions. Moving beyond one-off scripts, this new, reusable mechanism leverages the Search API to dynamically identify specific asset groups.
Dynamic Filtering: Use existing Search API criteria to select matching assets or story variants.
Bulk Execution: ability to perform mass actions, such as bulk publishing or unpublishing, across filtered sets of content.
Enhanced Data Governance & Security: To protect against accidental mass changes, we have implemented rigorous security standards and data governance. These safeguards ensure that bulk operations are executed intentionally and only on the intended groups of assets.
Refining Data Governance: Reports Retention Policy
To maintain peak system performance and storage compliance, we have implemented a standardized Reports Retention Policy. This development introduces an automated lifecycle for generated data. It also ensures that editors always have a clear way to retrieve the information they need.
30-Day Lifecycle: Generated report files, including Analytics, Assets (product-based and story-based), and other exportable types, are now automatically removed from storage 30 days after creation.
Data Integrity and UI Feedback:
Expired Downloads: In the Report Log, once a report exceeds the 30-day retention period, the “Download” button is deactivated. A tooltip informs the user: “The report has expired. Please check report details”.
Seamless Regeneration: The underlying source data remains intact in the database. From the “Report details” page, users can easily click the1 Regenerate button to generate a new report using the exact same filters as the original.
Preserved Filters: All original parameters – such as date ranges, story variants, and languages – are saved, allowing for consistent data retrieval even after the initial file has expired.
AI Chat (Alpha) Optimizations
The AI-driven image generation workspace has been upgraded to improve reliability and support a more flexible multimodal architecture.
New Interaction Type: Background Editing
Previously, the AI Assistant was focused primarily on generating new backgrounds from scratch. In Sprint 93, we are introducing a more precise way to manipulate visuals.
From Generation to Iteration: Editors can now use the new Image_edit prompt type to modify existing or previously uploaded backgrounds.
Creative Control: Users are no longer limited to “starting over.” You can now upload a background and ask the AI to add specific details – such as holiday-themed elements or festive decor – or even add and change objects within the scene.
Precision Adjustments: This interaction type allows for fine-tuning the composition while preserving the core layout, making it easier to adapt a single asset for different seasonal campaigns or regional requirements.
Othe updates and optimizations:
Vertex AI Migration: We transitioned from the public Gemini API to Vertex AI (via GCP service account) to reduce errors and improve connection stability.
Multi-Provider Support: The system now supports a multiple-client setup. We integrated OpenAI (gpt-5-mini + gpt-image-1.5) to support multi-turn chat dialogues and added support for Gemini 2.5 Flash Image – the first model in the “Nano Banana” series.
Performance & Output: Canvas exports now use WebP instead of PNG for better compression. We also implemented per-attempt timeouts, exponential backoff retries, and dimensions constraints for product references to prevent upscaling issues.
Refined UX: Added Edit and Copy actions for user messages, and a Retry button for failed provider responses. The preview modal now features a primary “Set as background” CTA for instant application.
Advancing Editor Efficiency: Story Builder v2
This sprint completes a major phase of our sidebar migration. It brings core module management directly into the host application, creating a smoother and more accessible experience.
Centralized Module Management
New Module Tree: The left sidebar now features a Tree view providing a visual overview of the story structure.
Direct Interaction: Editors can reorder modules via drag-and-drop and execute actions – like duplication, deletion (with confirmation), and styling – directly from the sidebar.
Bidirectional Sync: We’ve established a seamless link between the host and the preview iframe. Selecting a component in the Tree automatically scrolls the preview to that element. Clicking a module in the preview highlights it in the sidebar.
Smarter Component Organization
Enhanced Filtering: Component groups now support multiple tags and taxonomies, enabling editors to filter modules by categories like “Video”.
Personal Favorites: Users can now save frequently used components to a personal favorites list for faster access during the build process.
Component Groups updates – Keywords; Favorites
We’ve significantly upgraded how editors interact with the module library. By moving component groups to a dynamic MongoDB collection, we’ve laid the groundwork for a more flexible asset management system. It also improves searchability across components.
Taxonomy & Keyword Search
To help editors find the right tools for their stories faster, we have introduced support for multiple tags and taxonomies within component groups.
Multi-Tag Support: Components can now store and return multiple tags via the API, allowing for more granular categorization. The Story Builder v2 left sidebar will use these tags to organize content. Editors can filter the module library by specific categories, such as “Video” or “Images.
Dynamic Metadata: Support for taxonomy search keys has been integrated into the API, ensuring that create, update, and read operations all support this new metadata layer.
Personal User Favorites
Recognizing that many editors rely on a core set of “go-to” modules, we have implemented a per-user favorites system.
Customized Library: Within Story Builder v2 sers will be able to save specific components to a personalized favorites list for near-instant access.
User-Level Isolation: Favorites are stored separately for each user, ensuring a tailored workspace that doesn’t interfere with other team members’ preferences.
API-Driven Management: The backend now supports full add, remove, and list functionality for favorite components, with built-in validation to ensure only existing, valid component IDs are referenced.
Product Story Framework v2: Architectural Modernization and Encapsulation
Sprint 93 marks the near-completion of our Product Story Framework v2 rewrite, moving toward a more stable, high-performance, and scalable architecture for future content delivery. This overhaul transitions our rendering engine to modern web standards, ensuring seamless integration across diverse retailer environments.
Shadow DOM & Web Component Standards
The core of Framework v2 is built on the Shadow DOM and custom element standards.
Complete Encapsulation: By utilizing a “shadow root,” the Product Story functions as a fully encapsulated web component.
Style Integrity: Our styles are isolated from the host website; Icecat Studio content cannot break the retailer’s CSS, and conversely, the retailer’s global styles cannot interfere with our layout.
Modern Logic: This approach allows all story logic to be self-contained, ensuring consistent behavior across client-side environments.
Modernized Package Ecosystem
We have moved away from “hard-locked” legacy packages and rigid prefix requirements.
Dependency Refresh: All internal packages have been updated to their latest versions, removing technical debt and improving security.
Conflict Resolution: Removing strict prefix bindings enables more flexible deployments and easier maintenance of the codebase.
Custom Lightweight Video Player
To optimize loading speeds and Core Web Vitals, we have replaced third-party video solutions with a proprietary player.
Built from Scratch: The new video player is written specifically for Product Story requirements, avoiding the overhead of general-purpose libraries.
Reduced Bundle Size: The custom player results in a significantly lighter JavaScript bundle, leading to faster page load times and improved performance on mobile devices.
Current Status
The framework is currently undergoing final fine-tuning and testing. Once these quality benchmarks are met, Framework v2 will be fully integrated into the Story Builder, becoming the new standard for all generated content
If you are interested, click here and book a demo with Icecat Studio.