Drives deeper viewer engagement with custom visual assets using Veo
Creates relatable messaging across regions with localization in over 140 languages
Synthesia uses Veo 3, Gemini, and Nano Banana to enable businesses to generate custom video content.
Synthesia uses Veo 3, Gemini, and Nano Banana to enable businesses to generate custom video content.
Simplifying enterprise video production with the latest AI models
For global businesses, video is one of the most effective ways to share knowledge, yet production often remains a complex process. Physical cameras, expensive reshoots, and the difficulty of updating content quickly mean that valuable information—such as safety training, sales enablement, and marketing content—can be slow to reach its audience. With its all-in-one AI video platform, Synthesia helps solve these problems by making video creation as simple as typing a script, allowing organizations to communicate more effectively at a lower cost.
Synthesia’s pioneering digital avatars reproduce the subtle nuances of human movement, resulting in natural, lifelike delivery for effective communication. However, while the avatars are highly expressive, the surrounding visual context previously relied on generic stock footage. Users were limited to a fixed library of backgrounds that didn’t always match their brand or the specific details of their scripts. To solve this challenge and enable users to personalize their videos, Synthesia turned to generative AI.
With new AI models being released every few weeks, engineering teams are often required to rewrite APIs and reconfigure workflows to keep integrating the latest tools into their applications.
At Synthesia, development speed is very important, so staying on top of the latest technology is critical. The fact that Google’s AI models are constantly evolving means we can stand on the shoulders of giants and focus on developing the latest capabilities for our customers.
Peter Hill
Chief Technical Officer, Synthesia
Keen to avoid this engineering burden, Synthesia needed a stable AI ecosystem with a wide range of AI models that would remain at the forefront of technology even as new models emerged. Synthesia chose Google Cloud because it manages the updates and improvements to its models directly. This meant the company wouldn’t need to keep switching providers or reconfiguring workflows to keep up with the latest standards.
“At Synthesia, development speed is very important, so staying on top of the latest technology is critical,” says Peter Hill, Chief Technical Officer at Synthesia. “The fact that Google’s AI models are constantly evolving means we can stand on the shoulders of giants and focus on developing the latest capabilities for our customers.”
Fast, creative customization with Veo and Nano Banana
By integrating a range of Google AI models through Gemini Enterprise Agent Platform, Synthesia can give its users more freedom to personalize their videos to their requirements. With Veo, users can now use simple prompts to create custom visual assets from scratch, meaning they’re no longer limited by stock footage archives. For example, users can generate specific imagery for its B-roll footage—such as a factory floor or a specific office setting—that blends naturally with the avatar footage to create a professional result.
Chirp 3 and Studio voices are by far the most accurate models we’ve used. They’re the best at understanding intent, speaker identity, context, and all those details we need to localize content with a level of precision that makes the final video feel truly native.
Peter Hill
Chief Technical Officer, Synthesia
Synthesia is also using Nano Banana for automatic data visualization. If a script mentions financial figures or sales growth, for example, the platform can automatically generate accurate infographics or charts to accompany the speaker. For its own research and development, Synthesia uses Gemini 3 through Gemini Enterprise Agent Platform for large-scale video annotation and tagging. Gemini helps the team analyze thousands of hours of video to identify attributes like multi-speaker detection and setting. By mapping real-world human behaviors and expressions from this footage, Synthesia can train its avatars to replicate those same subtle nuances, making the digital speakers even more lifelike.
For global organizations, Chirp 3 and Studio voices within Cloud Text-to-Speech, along with the Cloud Translation API, enable localization in over 140 languages. This process preserves speaker identity by maintaining the unique vocal tone and character of the original person in every translated version, ensuring the message feels personal and recognizable across different regions.
“Chirp 3 and Studio voices are by far the most accurate models we’ve used,” says Hill. “They’re the best at understanding intent, speaker identity, context, and all those details we need to localize content with a level of precision that makes the final video feel truly native.”
Transforming enterprise training with two-way interactive media
Synthesia has seen a rapid increase in how global companies use its platform, moving from initial testing to large-scale adoption. In just seven months, the Synthesia community has produced 574 hours of video, which is the equivalent of 24 days of continuous playback.
These figures represent a fundamental change in how companies communicate: with over 15,000 Veo-powered videos generated weekly, AI-generated video is shifting from a novelty to an essential tool for training and internal updates.
The company is now moving beyond one-way broadcasts with plans to develop two-way interactive media. This evolution will involve leveraging Gemini Live API to build a skills platform designed for real-time behavioral learning. This will allow employees, such as sales representatives, to practice their pitches with interactive AI agents that can respond to follow-up questions and handle objections in a natural way. By focusing its engineering efforts on these new features, Synthesia is changing how people interact with digital content. For Hill, these future developments have been made possible also thanks to the company’s work with Google Cloud.
By integrating Google AI models with our own technology, customers can now create assets that are more dynamic, more engaging, and more relevant than they would have been previously. Working with Google models has been a gamechanger for our customers.
Peter Hill
Chief Technical Officer, Synthesia
“By integrating Google AI models with our own technology, customers can now create assets that are more dynamic, more engaging, and more relevant than they would have been previously,” Hill concludes. “Working with Google Cloud has been a gamechanger for our customers.”
Synthesia is an AI video platform that enables businesses to create professional videos with AI avatars in minutes, saving 90% in costs compared to traditional production methods.
Industries: Professional Services, Technology
Location: UK
Products: Google Cloud, Cloud Text-to-Speech, Cloud Translation API, Gemini Enterprise Agent Platform, Nano Banana, Veo 3
