Storyblocks, an unlimited, subscription-based stock media platform with video creation and workflow functionality, has unveiled new search enhancements powered by AI. These improvements aim to deliver an elevated user experience and streamlined creative workflows for Storyblocks’ customers.
In the recently released 10th Annual Survey of Creative Pros, Leslie Hughes, founder of iSPY Technologies and president of the Digital Media Licensing Association, said, “This year’s survey confirmed what creators truly value when licensing stock media – high content quality and relevance. However, finding the perfect assets through search remains their biggest hurdle. Encouragingly, many creators are optimistic that AI can lend a hand in streamlining asset discovery going forward.”
“Our team has relied on Storyblocks for several years now to create video content at a rapid pace,” said Nolan Nichols, manager of video production at the U.S. Travel Association. “We’re excited that Storyblocks is leveraging AI more to help customers like us find the content we need even quicker than before. We’re a lean team and are always looking for ways to work smarter, faster, and more efficiently, so I’ve been actively looking for ways that AI can help us do that at every step of the creative process.”
Cutting-edge data, AI and machine learning technologies drive Storyblocks’ platform. The company uses advanced AI models to deliver the most relevant content to its users. However, Storyblocks is not just about technological prowess; it is equally committed to curating a diverse and representative library that meets changing customer demands.
Also Read: Cision and Reddit Announce Expanded Partnership to Deliver Richer Social Insights
Recent improvements include:
- Delivering faster, more relevant and more representative results in search queries
- Storyblocks prioritizes fresher content in its search results through a proprietary algorithm that evaluates the performance of new content every four hours. Early testing indicates that including recently uploaded content can boost download rates by up to 60% compared to searches without this prioritization.
- Storyblocks employs an in-house machine learning model based on the BERT framework, allowing the platform to rapidly analyze how users engage with content and refine search result rankings in near real-time.
- Storyblocks aims to provide representative search results. On average, 24% of videos on the first page feature Black, Indigenous, or People of Color (BIPOC) models. This aligns with the company’s goal of proactively surfacing the quarter of its library content featuring models from underrepresented backgrounds.
- Furthermore, Storyblocks stands out in the stock media industry by routinely pulling underperforming assets to maintain a top-tier library quality. Despite expanding its collection by 40% in 2023, the company prioritized removing nearly 200,000 assets, ensuring customers can discover relevant content more efficiently.
SOURCE: PRNewswire