Overview

MLSE (Maple Leaf Sports & Entertainment) wanted to create a deeper connection between fans and the Toronto Maple Leafs through an innovative digital experience. Their solution? A real-time AI avatar generator that allowed fans to transform themselves into custom avatars, celebrating their love for hockey.

To bring this vision to life, MLSE developed its own ComfyUI-based AI workflow but needed a flexible, production-ready infrastructure to iterate quickly, deploy at scale, and meet strict enterprise requirements. That’s where InstaSD came in.

The Challenge: Scaling AI Avatar Generation for Enterprise Use

While MLSE had developed the AI workflow in-house, deploying it in a high-traffic production environment posed several technical hurdles:

✔ Need for Rapid Iteration and Customization

MLSE required a flexible environment to refine their AI pipeline with custom-trained models and easily push new versions to production. Existing solutions lacked the agility needed for continuous improvement.

✔ Scalability and API Deployment

Handling hundreds of avatar requests per minute with low latency was crucial for both the mobile app and live event activations. MLSE needed a scalable AI API deployment that could support unpredictable spikes in demand.

✔ Enterprise Compliance and Cost Control

As a major sports and entertainment company, MLSE required strict data location isolation, retention policies, SLA agreements, and most importantly, a fixed cost per inference to manage their budget effectively.

Other hosting providers lacked the agility, privacy controls, and scalability required to meet these demands.

The Solution: InstaSD’s ComfyUI Cloud & Scalable AI API

MLSE turned to InstaSD’s ComfyUI API deployment to power their AI avatar generator. Our platform provided:

🚀 Rapid Development with Scalable AI Infrastructure

With our ready-to-go infrastructure, MLSE’s AI team was able to focus on refining their workflow and AI pipeline without having to worry about complicated deployments, infrastructure setup, and scaling.

💡Instant Deployment to a Production-Ready AI API

Once finalized, MLSE seamlessly deployed their workflow to InstaSD’s serverless AI API, integrating it into their mobile app using our REST API.

⚡ Optimized Performance & Scalable AI Inference

To handle high event traffic and real-time avatar processing, InstaSD provided:

• Optimized GPU selection to reduce inference time.
• Parallel processing to handle hundreds of avatar requests simultaneously.
• Fixed cost per inference, allowing MLSE to budget for their expected volume.
• Serverless auto-scaling API deployments, which allowed MLSE to handle large user spikes during promotions and live events. With our setup, they only paid for what they used, avoiding unnecessary costs when the avatar generator was not in use.

🔒 Enterprise-Grade Customization & Compliance

Our team worked closely with MLSE to meet enterprise-specific needs, including:

• Custom data retention policies to align with privacy regulations.
• SLA-backed guarantees for uptime and performance.
• Flexible contract options to accommodate MLSE’s operational needs.

The Results: Driving Engagement with AI-Powered Experiences

Increased Fan Engagement

• Thousands of fans participated in the activation at NextGen games, engaging with the AI-powered booth.
• Seamless integration into the MLSE mobile app, tying avatar generation to a promotional ticket raffle.

High-Performance AI Deployment

• Real-time AI inference ensured instant avatar generation with minimal latency.
• Parallel request handling enabled high concurrency without performance degradation.
• Cost-efficient scaling, ensuring MLSE could handle peak traffic without overspending.

By leveraging InstaSD’s ComfyUI API deployment, MLSE successfully created an immersive, AI-driven fan experience—bringing cutting-edge generative AI into professional sports engagement.

‍

Conclusion: AI at the Heart of Fan Engagement

MLSE’s success highlights the power of flexible AI deployment. With InstaSD, they could:

• Build and iterate without infrastructure headaches.
• Deploy their AI workflow as a scalable AI API.
• Meet enterprise compliance standards seamlessly.
• Predict costs with a fixed price per inference, ensuring budget control.
• Leverage auto-scaling serverless deployment to handle variable traffic patterns efficiently.