A leading premium wildlife stock footage platform came to Armakuni for aI-native platform modernization: species detection, automated metadata, agentic content intelligence, full AWS rebuild.
The challenge
A species detection PoC in September 2025 grew into full platform ownership across six SOWs in five months. Armakuni now owns the architecture of the entire A leading premium wildlife stock footage platform platform.
Scale: 750+ videographers, 5,000+ hours of footage, 2,500+ species, serving filmmakers, broadcasters, and global brands.
What we built
Armakuni delivered aI-native platform modernization: species detection, automated metadata, agentic content intelligence, full AWS rebuild. The engagement ran as species Detection PoC, Content Ecosystem Assessment, Agentic AI Workflows, Content Ecosystem Modernisation, AWS EBA, DevOps Care Pod. AWS funded the work at $25K.
The outcomes
7 measurable outcomes shipped across the engagement. The ones that moved the business the most:
5,000+ hours of footage are now within reach of AI-automated metadata generation. This replaces a manual workflow where contributor tagging was the only path to publication, so footage that could never be submitted before can now be processed.
2,500+ wildlife species are searchable by behavior, habitat, and visual context. Buyers describe what they need and the system finds it, and the same pipeline scales to the full species inventory without architectural change.
10 TB single-file uploads are supported across 50+ professional video formats. With seven preview assets generated per upload, a constrained manual submission process became a production-grade ingestion pipeline handling every camera format contributors use in the field.
GPU-accelerated transcoding auto-scales with response under three minutes. The pipeline absorbs demand spikes that previously caused silent failures, processing RED and ProRes files through an event-driven architecture with full visibility at every job stage.
Metadata tagging reaches 95%+ accuracy across the entire library. Species identification, behavioral annotation, cinematic quality scoring, and habitat classification build a structured, high-quality metadata foundation with no human annotator in the loop.
A new dataset licensing revenue stream opened through 5-second behavioral segmentation. Scientific institutions and AI training teams can now license precision behavioral footage at a granularity the library could not previously offer, a market that did not exist before the platform.
AWS Marketplace listing capability was live from launch. The platform can reach enterprise buyers and global distribution channels without a separate integration program, with billing, subscription management, and private offers all in place.
Built on AWS
The production environment runs on 18 first-party AWS services. Delivered under our AWS Migration and Modernization, AI Services competencies.
What's next
Full rollout across 2,500+ species, AWS Marketplace listing, scaling agentic workflows across the complete video library.