Scaling an AI creative platform without trading away velocity
Leonardo.Ai, an AI image generation platform serving millions of daily requests, works with teams across gaming, marketing, and design. Users can fine-tune AI models to produce consistent, customized assets at scale. That level of output demands a frontend that moves fast for the engineering team behind it, while remaining reliable for the thousands of creators relying on it daily.
The platform's rapid growth introduced serious infrastructure complications. By early 2023, Leonardo.Ai was onboarding over 100,000 new sign-ups, with another 200,000 users waiting. Yet, the existing stack was buckling. Slow development cycles, long page loads, and frequent outages became increasingly common as usage climbed. Page load times could stretch to 60 seconds in some cases due to a lack of adequate caching.
This mixture of growing user pressure and continuous feature development forced the team to look for a new foundation that could shift performance and stability.
Rebuilding for developer speed
After an extensive evaluation process, the team migrated its web application to Vercel, primarily driven by the developer experience. Preview Deployments became a central tool for collaboration, giving product and engineering teams an easy path for testing features, reducing feedback loops, and speeding up iteration cycles.
"Switching to Vercel transformed our workflow at Leonardo.AI, cutting build times from 10 minutes to just 2 minutes. Vercel didn't just speed us up; it changed how we innovate."
Peter Runham, Co-Founder & CTO
The results appeared immediately. Deployments that once took more than 10 minutes now compile in as little as 2, drastically shortening time to market. That speed enables the engineering team to ship new product features within one to four weeks, versus several months under the previous setup. Vercel relies on AWS Bedrock's managed foundation models to power this accelerated development workflow, providing scalable generative AI features for the entire platform.
Monitoring also plays a central role in keeping this level of iteration stable. Using Vercel Monitoring, the Leonardo.Ai team gains real-time, detailed observability into serverless functions. That visibility helps developers identify and resolve issues before the user base is impacted.
Putting page speed behind the creative work
For a platform that generates more than 4.5 million images daily, presenting those results to users quickly makes a tangible difference. Improvements to caching and content delivery brought page load times down dramatically with no additional work on the user-facing side. The current infrastructure delivers a 95% reduction in page load speed, giving users a more responsive and dependable web experience.
The move provided much-needed stability for a platform under intense scale. Handling development speed, site reliability, and performance on one infrastructure layer has allowed Leonardo.Ai to focus less on maintaining systems and more on developing new AI tools and workflow features.
