Kami Vision launched generative AI–powered home security solutions in one month using Amazon Nova Lite

To enhance user experience and product intelligence, Kami Vision integrates generative AI into its camera solutions. This enables value-added services such as video question answering (VQA) and event notifications for scenarios including home security and elder care. The solutions deliver capabilities of intelligent video understanding and real-time response, helping Kami Vision win user trust and strengthen brand recognition.

To enhance the intelligent capabilities of its products, Kami Vision evaluated and compared multiple foundation models, and ultimately selected Amazon Nova Lite on Amazon Bedrock. Amazon Nova Lite delivers high cost-efficiency and outstanding performance in keyframe recognition and inference, making it well-suited for large-scale consumer-facing deployments. Compared with traditional methods that rely on extracting and analyzing individual video frames, this model directly analyzes video, eliminating the image conversion process. This significantly reduces inference latency and improves inference accuracy.

Using Amazon Nova Lite, Kami Vision has reduced token consumption per inference to 20% of the original and inference costs to 63% of the original, while maintaining a 94% video recognition accuracy, significantly improving the overall cost efficiency.

Amazon Nova Lite supports multimodal inputs, including images, text, and video, and offers the prompt optimization capability. The model maintains reliable recognition performance even on compressed or low-resolution images. Compared to models that can only process high-quality image inputs, Amazon Nova Lite offers greater performance and improved cost-efficiency. It also provides robust multilingual capabilities, aligning with Kami Vision’s global expansion goals.

Amazon Bedrock is a fully managed service that offers multiple high-performance foundation models with a single API, enabling enterprises to flexibly invoke different models and seamlessly integrate AI across various development stages. With Amazon Bedrock, Kami Vision can choose foundation models best suited for use cases of small and medium-sized enterprises and quickly deploy them to production. Flexible prompt configuration enables the product team to improve response accuracy and consistency during prototype verification and post-launch iterations.

AWS services and products help Kami Vision significantly accelerate the development cycle — from identifying business needs to delivering solutions. This reduces technical barriers to AI adoption, enabling faster deployment of generative AI and continuous innovation on user-facing services.

As Kami Vision expands globally, it faces a complex landscape of data privacy and processing regulations that vary across countries. Security products that handle sensitive data, such as video and images, must strictly comply with regional regulations to ensure that data processing remains secure, controlled, and compliant. In mature markets such as North America and Europe, stricter regulations on data processing and storage impose higher compliance requirements on security solutions.

AWS’s global infrastructure and regional compliance expertise help Kami Vision expand its international footprint and implement solution deployment in markets such as North America, Europe, and Southeast Asia. Amazon Bedrock provides robust compliance controls and detailed documents, offering Kami Vision clear governance guardrails for integrating generative AI and implementing responsible AI adoption at a global scale.