Our Platform Approach to AI Performance

A Flexible Foundation

We deploy a versatile AI computing platform rather than a single, fixed AI model or proprietary software solution. Our platform is built on well-established, frequently updated open-source software that serves as a foundation for running any compatible open-source AI model. This approach provides several key advantages:

Flexibility: Choose from thousands of available open-source models to find the perfect fit for your needs
Adaptability: Easily switch between different models as your requirements change
Future-Proofing: Take advantage of new models as they become available
Transparency: Benefit from continuous improvements in the open-source AI community
Control: Fine-tune the balance between model sophistication and processing speed

The platform we deploy is actively maintained by a global community of developers and researchers, ensuring regular updates, security patches, and new features. Rather than locking you into proprietary software, we provide a robust foundation that can evolve with your needs and the rapidly advancing AI landscape.

Understanding Our Demo Platform

Entry-Level Business GPU

Our on-premises AI demo runs on an AMD Radeon RX 7600 XT, a 16GB graphics processing unit (GPU) that we consider an entry-level platform for businesses. This GPU allows users to load and work with many commonly available AI models without being limited to tiny model sizes, making it a practical starting point for on-premises AI deployments. While there are lower-end GPU options available, we find they can be too limiting for the needs of most business customers. The RX 7600 XT represents a sweet spot of performance and memory capacity that enables a smooth, capable demo of what an entry-level on-premises AI setup can deliver.

Why We Chose This Configuration

Think of our demo platform as “Stage 1” in the AI computing spectrum – it’s carefully designed to give businesses a tangible feel for on-premises AI without overwhelming complexity or cost. This configuration helps answer the crucial question: “What does it feel like to run AI in my own building?” It provides a realistic experience of response times, capabilities, and limitations that you might encounter in a basic deployment.

Understanding AI Hardware Tiers

Entry Level (Our Demo Platform)

GPU Memory: 16GB
Suitable for: Small businesses, initial deployments, basic AI tasks
Capable of running: Most consumer-grade language models, basic image processing
Ideal for: Text generation, customer service bots, basic document analysis
Concurrent Users: 1-3 typical users

Mid-Range Solutions

GPU Memory: 24-32GB
Suitable for: Medium businesses, dedicated AI workflows
Capable of running: Larger language models, advanced image processing
Ideal for: Multiple concurrent AI tasks, faster processing times
Concurrent Users: 3-10 typical users

Power Users

GPU Memory: 48-80GB+ (often multiple GPUs)
Suitable for: Large organizations, intensive AI workloads
Capable of running: The largest available models, multiple models simultaneously
Ideal for: Research, complex analysis, high-volume processing
Concurrent Users: 10+ users

The Hardware-Model Trade-off

Understanding the relationship between hardware capabilities and AI model performance is crucial for making informed decisions about your AI infrastructure:

Model Size vs. GPU Memory

Larger models generally produce better results but require more GPU memory

A 16GB GPU (like our demo) can handle models up to about 13GB

Remaining memory is needed for processing and temporary storage

Multiple users require additional memory overhead

Scaling Your AI Infrastructure

For customers interested in exploring higher-performance hardware options for their on-premises AI projects, we encourage you to contact us to discuss custom arrangements and demonstrations. Our team is happy to work with you to identify the right hardware configuration to meet your specific requirements.

Making the Right Choice

When helping clients choose their AI hardware configuration, we consider several factors:

Expected number of concurrent users
Types of AI models needed
Response time requirements
Budget constraints
Growth projections
Security requirements

Our demo platform provides a baseline for understanding these considerations in practice. By experiencing an entry-level system firsthand, you can better evaluate how scaling up might benefit your specific use case.