Swann adds AWS generative AI to cut camera alert noise

Swann has introduced an artificial intelligence feature for its home security app that filters camera alerts based on user-defined scenarios, in a move aimed at reducing notification overload for customers.

The Notify Me When feature is now available in the Swann Security App. It uses generative AI models from Amazon Web Services to analyse footage from Swann cameras and send alerts only when events match specific prompts created by users.

The launch extends an existing system called Dynamic Notification. That system already sends alerts with thumbnail images and short descriptions of motion events.

Swann said the combined tools have cut false positive alerts by up to 65 per cent in its testing. The company also reported a 25 per cent drop in overall alert volume and an 89 per cent increase in notification relevance.

Swann is based in Melbourne and sells do-it-yourself home security products globally. The company developed the new alert features using AWS cloud services and AI tools.

Dynamic Notification activates when a Swann camera detects motion. The system then captures a still image and generates a short text description of what the camera saw.

Users receive that information as a push notification on their phone. They can review the image and description without opening the app or watching the full clip.

The Notify Me When function adds a further filter. It allows users to set custom prompts that describe the specific scenes or actions that warrant an alert.

Swann said examples include rules such as “notify me if a dog enters the backyard”, “notify me if a child is near the swimming pool” or “notify me when a package has arrived”. The system then checks each motion event against those prompts.

If the AI-generated description of a scene aligns with the user’s scenario, the app sends a notification. If it does not match, the system can suppress the alert.

The descriptive alerts use Amazon Bedrock and the Amazon Nova suite of AI foundation models from AWS. These services run in the cloud on AWS infrastructure.

When a Swann camera registers movement, the video is uploaded for analysis by an Amazon Nova model. The model interprets the visual content in the frame.

The system then converts that analysis into a short text statement. That statement describes the objects, people or actions that appear in the scene.

The Notify Me When feature compares this statement with the user’s prompt. It then triggers a push notification if the description meets the defined condition.

Swann framed the updates as a response to customer concerns about frequent and irrelevant alerts from connected devices. The company said some users risk missing important events when they face constant notifications.

“When it comes to security, we’ve heard feedback from many of our customers that too many false positives lead to alert fatigue, which can increase the risk of people turning off notifications or simply missing real security events that require swift responses,” said Surjit Reghunathan, Chief Technology Officer, Swann. “We’re proud to leverage some of the latest AI technologies from AWS to develop the Notify Me When feature and the Dynamic Notification tool to help customers keep the focus on real threats.”

AWS said Swann’s adoption of its AI models shows how generative AI is moving into consumer devices. The cloud provider has worked with Swann on several features in its product line.

“Swann’s innovation in smart security solutions reflect our combined commitment to customer obsession and innovation. By addressing customer feedback on alert fatigue challenges, Swann developed two AI-powered services that deliver intelligent alerts, which has significantly increased customer satisfaction in Swann’s smart security solutions,” said Nam Je Cho, Director for Solutions Architecture, AWS Australia and New Zealand. “Designed in Melbourne, Swann has expanded globally by leveraging AWS’s cloud infrastructure, helping them deliver superior performance and reliability to customers around the world. It’s great to see practical applications of generative AI moving beyond chatbots into everyday consumer devices, helping people in their daily lives, especially through multi-modal AI capabilities.”

The launch follows Swann’s earlier release of SwannShield in 2024. That service uses Amazon Bedrock as the basis for a voice assistant that responds to visitors in over 100 auto-detected languages.

Swann markets its products through major retailers and distributors in North America, Europe, Asia and Australia. The company focuses on do-it-yourself security systems for homes and small businesses.

Swann said it will continue to build AI-based features on AWS services as it develops new products and updates for its security portfolio.