Challenge
A PE-backed global automotive SaaS company serves dealership networks across multiple international markets, processing high volumes of customer-facing sales and workshop videos every day. At that scale, manual review became the bottleneck. Reviewing thousands of videos by hand was slow, inconsistent between reviewers, and difficult to scale as volume grew.
Maintaining a consistent quality standard across dealership networks in different markets added another layer of difficulty, since standards applied by one reviewer rarely matched another exactly. On top of that, the platform had limited ability to automatically surface the customer-relevant moments buried inside long videos, which meant genuinely useful content, a clear vehicle feature, a walkthrough moment, a compliance issue, often went unseen simply because nobody had time to find it.
Solution
NexaQuanta built an AI video quality scoring engine powered by vision-language models and computer vision, automating video assessment, highlight generation, timestamping, and confidence scoring across the platform. The system runs on a scalable, cloud-native processing architecture with API integration, so it fits directly into the client’s existing workflow rather than requiring a separate review step.
The stack combines OpenAI LLMs and vision-language models for understanding video content, computer vision and YOLO for object detection, and speech-to-text for spoken content, all running on a queue-based distributed processing architecture in AWS, with FastAPI microservices and New Relic monitoring keeping the pipeline reliable at volume.
Results
$6M in additional revenue unlocked for the business.
- Consistent video quality assessment across the entire global dealership network, replacing reviewer-by-reviewer variation with a single standard.
- Every dealer video scored and highlighted automatically at the point of upload, at a task that previously required a human editor.
- Improved customer engagement through AI-generated highlight reels, and faster identification of vehicle findings and upsell opportunities at scale.
What this means going forward
This wasn’t a quality-scoring tool bolted on as an afterthought. It runs natively inside the platform, at the point of upload, across every market the client operates in. For a SaaS company competing on the strength of its video content, that’s the distinction that matters: not whether AI can score a video, but whether it can do so consistently enough, and fast enough, to become a genuine revenue driver rather than a cost centre.
Ready to see what’s possible in your software?
Explore our Agentic SaaS Solutions →
