The Problem

High costs and rigid infrastructure slowed growth.

Gendo is a browser-based AI platform that enables architects and designers to produce high-quality CGI images of their designs in minutes, significantly reducing the time and resources traditionally required for architectural visualization.

However, as Gendo experienced rapid growth, their existing AWS infrastructure became a bottleneck. The setup lacked scalability, making it difficult to handle varying workloads efficiently. Developers found themselves dedicating excessive time to infrastructure and DevOps tasks, detracting from product development. The limitations became particularly evident during beta testing, where scaling GPU resources to meet user demand proved problematic.

"Before using RunPod, we struggled with our AWS infrastructure. Setup wasn't particularly scalable, and we struggled to go from zero instances up to many. We didn't always have the resources on demand in the way that we wanted.”

Additionally, the team was spending more time managing infrastructure than focusing on product features:

"One issue that we faced was that developers were spending too much time on infrastructure and DevOps. Instead of spending time on product features, we were spending time building this GPU backend.”

The Solution

Flexible, Autoscaling GPUs with Serverless

To overcome these challenges, Gendo integrated RunPod’s dynamic and flexible GPU infrastructure, allowing them to optimize both cost efficiency and performance. By leveraging RunPod’s serverless offerings, Gendo could dynamically scale GPU resources up and down based on real-time demand, eliminating wasteful spending on idle GPUs.

“We discovered RunPod over a year ago, and to start with, we were just using the pods feature for more ad hoc GPU tasks. Then I noticed that RunPod really was serverless, and that was a big thing for us because it solved a lot of the scalability issues that we were having within our own infrastructure.”

The migration to RunPod was seamless, thanks to Gendo's use of containerized images:

"Moving to RunPod was really smooth. We were using containerized images before, and because RunPod allows us to point to those same images, we were able to move across really easily.”

The Results

Lower Costs, Faster Performance, and a Seamless User Experience

By switching to RunPod, Gendo achieved substantial cost savings while delivering a smoother and more responsive user experience. Their pay-per-job model eliminated unnecessary expenses, allowing them to invest more into product innovation rather than infrastructure overhead.

"The great thing about RunPod is that the pricing becomes a bit more predictable, and we have better unit economics as a business. Whereas if we rolled our own infrastructure, the costs are much more unpredictable, and we end up spending more inefficiently.”

With RunPod’s smart scaling, Gendo ensured that architects and designers could generate high-quality visualizations without delays, even during peak demand:

"Since switching to RunPod, scalability has been solved in a way that we couldn't do before. We had a very fixed infrastructure, and quickly starting up new containers and new ML services was slow. With RunPod, that process is much faster.”

‍

The improved reliability and efficiency of their platform strengthened customer satisfaction and positioned Gendo as a leader in AI-driven architectural visualization.

"RunPod has allowed the team to focus more on the features that are core to our product and that are within our skill set, rather than spending time focusing on infrastructure, which can sometimes be a bit of a distraction.”

Conclusion

By integrating RunPod into their operations, Gendo overcame critical infrastructure challenges, leading to improved scalability, performance, and cost efficiency. This partnership allowed Gendo to deliver on its mission of providing architects and designers with rapid, high-quality visualizations, fundamentally transforming the architectural visualization landscape.

"RunPod is a great platform for founders like me who want to ship AI models to customers quickly and reliably.”