The problem
Qatar Tourism had accumulated thousands of hours of high-quality tourism footage across attractions, experiences, and destinations. Finding specific content within that library required manual searching through metadata and tags — slow, inconsistent, and dependent on how well footage was originally labelled.
What we built
An AI video processing and search system that analyses footage at the frame level — identifying locations, activities, people, objects, and visual themes. The system creates rich searchable metadata automatically and powers a natural language search interface. ‘Show me footage of desert experiences at sunset with families’ returns relevant clips instantly.
What changed
Thousands of hours of tourism footage became a searchable, discoverable asset library. Content that was previously invisible — because it was buried in an unlabelled archive — is now findable and usable for campaigns, social media, and media requests.
The honest part
Video AI is compute-intensive. Processing a large existing library requires upfront investment in time and infrastructure. The ongoing value is high, but the initial processing run needs to be planned and budgeted for.
Working on something similar?
Tell us what you're trying to solve. We'll give you an honest answer.