Generative AI models have reached a baseline ability to generate at least passable videos from a single image or short sentence. Companies developing products based on these models claim that anyone with images or recordings can create stylish promotional videos. Also, videos usually perform better than still images or documents.
Avatar, backed by Peak XV and Tiger Global, released a new tool on Monday called Velocity. Create product videos directly based on product links. The company will compete with companies like Amazon and Google, which are also experimenting with AI-powered video tools for advertising.
The startup ethos behind these tools is similar to video creation tools like Lica and Sythesia. Videos are expensive to create, so companies can't create them all. But thanks to AI, the cost of video production has come down and clips can now be generated in bulk.
Avatar was founded in 2015 by Sravanth Arul, a former Microsoft and Deutsche Bank employee. The company, which has raised more than $55 million in funding to date, previously focused on creating interactive experiences, AI-powered images, and 3D models for e-commerce. The startup is currently focused on using AI to create videos for brands.
The company says 3D models and experiences have higher conversion rates than videos. However, because far more people use videos, they have an advantage in terms of scale, even if the conversion rate is lower.
The company began commercial testing of its AI-powered video creation tool with customers in February 2024. But back then, Avatar still needed to involve people in the process of helping these brands create videos. The company began testing the fully automated Velocity tool in September 2024.
Its tools are already used by HP, Victoria's Secret, Lowe's, Newegg, Marina, TVS, and Bajaj to generate a variety of product videos.
While Avatar allows brands to use its services and tools to create customized videos for their flagship products, it also helps brands create product clips for products that aren't strategic enough to garner video production budgets. He said Velocity would come in handy. The startup hopes to improve conversion rates for these products through informative AI-generated videos.
“Through Velocity, we can cover our entire catalog and create videos for every item in our catalog. This improves the customer experience and also drives higher conversions through motion and storytelling around our products. ,” Aluru told TechCrunch.
The company also has an API that allows businesses to integrate with the platform to create product videos for automated listings.
Aluru said Avatar's models and solutions are different because they collect a lot of data about the product through 3D rendering. He noted that the brand cannot afford to have glitches in its product presentation videos, and the company is making sure to do so.
“Our model understands the different subcategories of products and their associated attributes. Our tool highlights different characteristics of the Hammer compared to a smartphone. We are focused on better storytelling using video as a medium, with the end goal being ,” he said.
Through videos, Avatar seeks to not only provide details to customers but also create aspirational value for the product.
The company noted that it has built guardrails around brand safety and product inaccuracies so that its models can check for inaccuracies in its creations and reject the output produced by its tools.
Amir Koenigsberg, founder and investor at defunct e-commerce startup Twigl, said there is an opportunity to create videos for online sales, but they need to be unique enough to stand out. Ta.
“Video generation from product links is a natural extension of e-commerce pipeline automation. Video has been proven to drive engagement and conversions, and it would be wise to scale this efficiently. However, Backgrounds, after effects, product renderings from different angles, and cutouts. There are already several companies working on aspects of this space, such as creating photorealistic shadows. The challenge here is to stand out, ensure consistency, and adhere to licensing and attribution standards. ” he told TechCrunch via email.
Konigsberg added that brands have not adopted AI-generated video solutions so far because it is difficult to maintain quality at scale. However, he hinted that recent advances in models could make these tools effective for mass use.