Enhances image generation precision and consistency across complex visual prompts with Nano Banana
Enables highly realistic video generation through Veo 3
Helps power 10x revenue growth and 3x user growth
Boosts productivity for business strategy and product development with Gemini
Using Google AI models, Lovart enhances its AI design agent’s image generation quality for smoother co-creation experiences.
Using Google AI models, Lovart enhances its AI design agent’s image generation quality for smoother co-creation experiences.
An AI creative partner that automates design workflows and sparks creativity
AI image generation has been increasingly integrated into design workflows to power productivity and creativity. However, for new users, it may be challenging to have their ideas accurately understood and executed by AI image generators.
To smooth human-machine co-creation experiences for all users, Lovart developed an AI design agent that acts like a human creative director. Launched in July 2025, its web-based AI agent can think, plan and execute multi-step design tasks, transforming and refining users’ ideas into professional-grade visuals with the maximum level of automation. For example, through the feature called ‘ChatCanvas,’ users can exchange their ideas with Lovart’s AI design agent directly next to the image under the design process.
Nano Banana excels in understanding visual and long context, resulting in highly consistent and precise image generation. It is exactly what we need to deliver high-quality AI-assisted design work.
Melvin Chen
CEO and Founder, Lovart
“With our AI design agent, we want to offer our users the same content creation experiences as with human designers. This means that our agent needs to be able to interact naturally, give inspiring feedback and deliver visuals as requested,” explains Melvin Chen, CEO and Founder at Lovart. “That’s why we’re constantly looking for the most advanced AI tools on the market to build our product.”
To achieve the highest image generation quality, Lovart leverages an agentic AI model and an image generation model to enable its image generation feature. The agentic AI model would first assist users with ideation, then generate task planning and prompts accordingly. Based on the prompts, the image generation model would create visuals. Since these prompts often contain images and long context, the image generator that Lovart used at first was not able to always comprehend them accurately, leading to inconsistent results. When Nano Banana of Google Cloud was launched, the Lovart team decided to incorporate it into its toolkit.
Enhancing AI image generation quality to unlock creativity with Nano Banana
Now powered by Nano Banana Pro, Lovart’s AI design agent is able to deliver images precisely meeting users’ needs with higher consistency. With Nano Banana Pro’s token input context window as large as one million, Lovart’s agent can execute complex design tasks while taking account of all the references and users’ preferences.
Nano Banana has helped us significantly expand our image generation capabilities, realizing features that facilitate human-machine co-creation and boost users’ productivity. Since we started using Nano Banana to support our AI design agent, our revenue has grown by 10x, and our user number has tripled.
Melvin Chen
CEO and Founder, Lovart
Besides, Nano Banana Pro’s capability to process visual prompts means that Lovart’s agent can better maintain style consistency based on the images given.
With the enhanced image generation capabilities, Lovart’s diverse image creation features now help users unlock their creativity more effectively. For example, one of its features called “Touch Edit” allows users to precisely add elements to designated areas. Prior to the adoption of Nano Banana, Lovart’s AI design agent sometimes struggled to perfectly execute such detailed commands.
On top of that, Lovart leverages Veo 3 to generate highly realistic videos. The capability of Veo 3 to fully synchronize image and audio, coupled with the succinct prompts generated by Lovart’s AI design agent, enables users to create high-quality videos for professional use in a short time.
Boosting business decision-making efficiency with Gemini
The Lovart team uses Gemini to facilitate decision-making in various operation aspects from pricing design to product development. Chen notes that Gemini’s outstanding context understanding and deep research capabilities make it a powerful tool to provide inspiration and assist with judgement.
“We’ve tested several large language models for business decision-making, and Gemini is the one that gives the most insightful suggestions and the most useful, up-to-date information,” he adds. “It serves as an excellent assistant that helps us make more timely and thoughtful business decisions.”
Leveraging Gemini’s multimodal understanding, Lovart also uses the model to tag users’ prompts and evaluate images generated by its AI design agent. This way, the team is able to analyze the performance of its products more efficiently, accelerating product improvement.
Lovart is currently exploring the possibility of using Gemini to support its agent’s agentic functions. Moving forward, the team plans to make its AI agent more proactive by suggesting relevant ideas or guidance ahead of requests. It will also leverage the multimodal capabilities of Google AI to gain better understanding of users’ visual perception and tastes to generate more diverse content.
To continue improving human-machine co-creation experiences, we want to better align the visual perception of humans and AI. This can only be achieved through more advanced multimodal comprehension of AI, and we believe that we can count on Google Cloud to reach this goal together.
Melvin Chen
CEO and Founder, Lovart
Lovart provides a web-based AI design agent that automates the entire creative process for image and video creation. It currently serves more than 5 million users around the world.
Industry: Technology
Location: China
Products: Google Cloud, Gemini, Nano Banana, Veo 3
