Case Study

How Digi Dash Was Built Using the Lid Vizion AI Integration Platform

Overview

Digi Dash is an interactive application built to demonstrate how AI, data, and custom logic can be combined into a production-style system. While the experience includes a game interface, the project’s real focus was on AI orchestration, asset generation, data management, and system configurability.

The application was built on the Lid Vizion AI integration platform and integrates multiple AI services, databases, and developer tools to reflect how real-world AI products are designed and deployed.

The Challenge

Building Digi Dash surfaced several challenges that are common in real AI products:

• Making AI outputs understandable to end users
  
  
• Reducing reliance on prompt engineering expertise
  
  
• Managing dynamic content and user state
  
  
• Supporting rich media, including 3D assets
  
  
• Allowing configuration without deep technical setup
  
  
• Maintaining security and deployment flexibility
  
  

The goal was to solve these challenges in a way that could be reused across other applications and industries.

Solution Architecture

AI Services & Capabilities

Text & Reasoning (OpenAI)

OpenAI was used to generate trivia questions and structured hints. One specific challenge was ensuring that hints were useful but not revealing.

To solve this, Lid Vizion implemented:

• Prompt templates with strict constraints
  
  
• Context-aware prompt generation
  
  
• Validation logic to ensure exactly two hints per question
  
  
• Difficulty calibration for knowledgeable users
  
  

This approach demonstrates how generative text models can be safely used in structured experiences such as training, education, or decision support.

Image Generation (Gemini)

Gemini image generation was integrated to dynamically create visual assets used throughout Digi Dash, including playful elements such as a banana.

This demonstrated:

• Programmatic asset creation
  
  
• Visual variability without manual design work
  
  
• Use of generative images in real applications
  
  

The same approach applies to marketing assets, UI visuals, and training materials.

3D Object Generation & Animation (Meshy)

Digi Dash also integrates Meshy for 3D object generation and animation.

Key Challenge

3D generation models typically require:

• Detailed prompts
  
  
• Technical configuration
  
  
• Trial-and-error prompt tuning
  
  

This creates friction for non-technical users.

Lid Vizion’s Approach

To reduce this complexity, the platform introduces an abstraction layer:

1. The system understands the context of the game or experience
   
   
2. That context is passed into an internal AI layer
   
   
3. The internal layer prepares and refines the prompt
   
   
4. The optimized prompt is then sent to Meshy
   
   

This means:

• Users do not need to focus on prompt engineering
  
  
• Dashboard users do not need to configure complex settings
  
  
• Generated 3D assets align with the experience automatically
  
  

This same pattern can be applied to other generative systems where prompt quality directly affects output quality.

Data & State Management

Flexible Data Modeling

Digi Dash uses a document-oriented data model to store:

• Trivia questions and answers
  
  
• Hint logic and constraints
  
  
• User sessions and progress
  
  
• Scoring and outcomes
  
  
• Dashboard configuration data
  
  

This flexible approach allows the system to evolve without rigid schema changes and supports both the interactive experience and administrative tooling.

Workflow Orchestration

Lid Vizion orchestrates multiple AI and logic steps within a single flow:

• User input handling
  
  
• AI generation
  
  
• Rule evaluation
  
  
• Feedback delivery
  
  
• State updates
  
  

These workflows are modular and reusable, allowing logic created for Digi Dash to be repurposed in other products.

Dashboard & Customization

A custom dashboard was built to manage:

• Content configuration
  
  
• Difficulty tuning
  
  
• Scoring rules
  
  
• Asset updates
  
  
• Game flow logic
  
  

This allows non-technical users to operate and evolve the system without redeploying code, which mirrors real enterprise needs.

Authentication & Security

Authentication with Auth0

Auth0 was used to handle authentication and user access.

This enables:

• Secure user login
  
  
• Role-based access control
  
  
• Separation between player and admin roles
  
  
• Compatibility with enterprise identity providers
  
  

Auth0 integration ensures the platform meets common security expectations while remaining flexible for different deployment environments.

Custom Models vs API-Based AI

During development, the team evaluated:

• Using existing AI APIs
• Training a custom trivia model

This highlighted a common tradeoff:

• API-based models enable faster iteration and lower upfront cost
• Custom models offer tighter control and domain specificity

Why a Custom Model Was Built

For trivia question generation, standard AI tools sometimes struggled with:

• Keeping difficulty levels consistent
• Formatting questions in a structured way
• Maintaining accuracy at scale

The Solution

The Lid Vizion team developed a custom trivia model trained on thousands of high-quality questions. This approach delivered:

• More consistent question difficulty
• Better formatting and accuracy
• Faster performance
• Lower costs at scale

A Flexible Platform

The custom model works alongside other AI services in the platform. Teams can:

• Switch between custom and ready-made AI solutions
• Test different approaches without changing code
• Choose the best option for each specific need

This flexibility means the platform can adapt as AI technology evolves, supporting both off-the-shelf APIs and custom models depending on what works best for each use case.

‍

Development Workflow

Modern AI-enabled development tools such as Cursor were used to accelerate coding, testing, and iteration. This allowed the team to:

• Rapidly prototype features
  
  
• Refine AI logic
  
  
• Maintain clean, reusable code

The workflow reflects how AI products are built in practice, not just as demos.

‍

Multi-lingual - API? Google Translate

Outcome

Digi Dash demonstrates how:

• Multiple AI services can work together in one system
  
  
• Prompt complexity can be abstracted away from users
  
  
• 3D assets can be generated and animated dynamically
  
  
• Dashboards can control AI behavior
  
  
• Security and deployment flexibility can be preserved
  
  

While presented as a game, the system reflects real product architecture and real AI integration challenges.

What This Case Study Shows

This project illustrates how an AI integration platform can:

• Reduce complexity for end users
  
  
• Enable faster experimentation
  
  
• Support rich, interactive experiences
  
  
• Scale across different applications and industries
  
  

Digi Dash serves as a concrete example of how AI components, data, and workflows can be combined into a working system