Launching a new business or sharing a personal passion online should be an exciting milestone, but for many individuals, the process can feel confusing and complex. Managing web development and setting up an online presence frequently require specialized technical skills that can stall great ideas before they get off the ground. Hostinger addresses this challenge with its accessible online platform designed to make digital success simple for creators, bloggers, and small business owners alike. To remove technical friction for its users, the company integrates artificial intelligence directly into its software, allowing creators to describe what they want in natural language to create complete web pages or promotional layouts.

However, as it scaled these conversational AI capabilities across 150 countries, Hostinger encountered operational issues, including high computing costs, slow response generations, and unpredictable outputs. When creators had to wait too long for results or received poorly formatted content, they would end up spending extra time manually correcting mistakes, causing them to lose trust in the automated tools. To protect the user experience, Hostinger required flexible AI models that could process complex prompts efficiently. The company selected Gemini models from Google Cloud, giving it the precise balance of quality, execution speed, and affordable, flexible model pricing required to handle complex conversational tasks.

“If an automated system takes too long to respond or provides inconsistent results, creators quickly look elsewhere,” explains Saulius Lazaravičius, vice president of product at Hostinger. “We chose Gemini models because they deliver the right balance of processing power and cost efficiency, giving us the capabilities to handle multi-step customer requests without unnecessary overhead.”

To deliver an immediate, hassle-free setup for its users, Hostinger embedded Gemini models into its key user-facing products. Inside Hostinger Horizons, a conversational tool where customers use chat to build complete websites and applications, the company built a dynamic model-switching architecture. This system automatically evaluates a user’s prompt to determine the complexity of the request and routes the task to the most appropriate system. Lighter requests like basic design updates or text corrections are handled by the faster Gemini Flash models, whereas deeper programming fixes or full-page generation tasks trigger the more advanced Gemini 3.1 model. By managing model selection intelligently in the background, Hostinger cut total application response times by 25% and reduced background error-checking windows from 40 seconds to just 12 seconds. Within this same workflow, Horizons uses the Nano Banana model to generate and edit custom web graphics. This delivers clean, fast image outputs at an affordable price point, ensuring website visuals remain polished without slowing down the initial page build.

For its automated email marketing tool, Hostinger Reach, the engineering team applied a different technical approach to balance performance with budget. Because drafting newsletters and promotional campaigns involves less intensive programming structures than building complete software applications, Hostinger didn’t need to run expensive, premium systems for every prompt. Instead, the tool combines Gemini 3.0 Flash and Gemini 3.1 Flash Lite with highly detailed, structured prompt guidelines. This precise configuration allows the smaller models to interpret user intent with a high degree of accuracy, automatically writing copy and building well-formatted email code while keeping costs low.

“Matching the technical complexity of a user request to the right size model has transformed how we manage our applications,” explains Lazaravičius. “By running less demanding layout generations on Gemini Flash models, we halved our operational costs and response latency while maintaining excellent visual standards.”

Focusing on user needs and hiding technical configuration behind automated conversational tools has allowed Hostinger to expand its global customer base significantly. The platform now supports over five million users worldwide, making it simple for people without a background in software engineering to create their desired digital projects. By matching distinct user tasks to specific generative options, Hostinger achieved clear, measurable business improvements. Transitioning its email automation tool to efficient Flash options cut response speed and operational expenses by approximately half, while the smart model-switching design inside its website creation tool lowered total wait times by a quarter. These optimizations allow everyday creators to launch complete promotional campaigns and custom websites quickly, saving them from the traditional costs of hiring outside agencies or developers.

Moving forward, Hostinger plans to continue making generative models more practical, responsive, and seamless for its global audience. As consumer expectations for conversational software grow, the company intends to use its strong relationship with the Google Cloud team to test next-generation systems and early model variants before they reach the wider market.

“With the help of Gemini models our users can bypass AI complexity to generate websites and email campaigns effortlessly,” concludes Saulius Lazaravičius, Vice President of Product at Hostinger. “The result is faster creation, better efficiency, and a smoother path for our users to succeed online.”