Qualcomm Insight Platform embeds native AI directly into the video stack, turning raw footage into a continuously queryable real-time intelligence layer.
This fundamentally changes the way we use video for safety, security, and operational efficiency.
Our edge-first architecture runs AI inference on-device or on local edge appliances, keeping raw video securely on-site. This design optimizes privacy and bandwidth by sending only metadata and specific events upstream.
Merlin is a built-in generative AI agent that provides an intuitive human language interface for querying video streams. This capability enables the platform to provide deeper insights and flexible deployment options beyond traditional VMS and bolt-on analytics.
Although video has emerged as one of the most information-dense data sources within the enterprise, most systems still treat video as a passive record that is stored, scrubbed, and archived. To address this, the Qualcomm Insight Platform makes AI a native feature of the video stack, transforming video from “footage” to a continuously queryable, real-time intelligence layer focused on safety, security, and operations. Designed around edge AI and profile-aware generative intelligence, Qualcomm Insight goes far beyond traditional VMS and bolt-on AI analytics.
Additionally, with the acquisition of Augentix, a provider of low-power ISP and multimedia SoCs, Qualcomm Insight is poised to offer a broader and more flexible portfolio of smart cameras, allowing system designers to optimize camera selection by zone while maintaining unified control and cost efficiency.
Interested in the Qualcomm Insight platform?
Book a demo now
Introducing the Qualcomm Insight Platform for VSaaS
Native AI, not add-ons
Most incumbents employ traditional video management solution (VMS) architectures with AI “add-ons” layered on top, often as separate services that are loosely coupled, difficult to scale, and have limited functionality. Qualcomm Insight is built around native AI. AI models are built into cameras, edge boxes, and control planes, so video understanding is a first-class capability, not an afterthought.
At its core, Qualcomm Insight fuses high-performance vision models (convolutional neural networks and transformer-based detectors) with vision language models (VLMs). This allows the platform to translate human language into structured queries according to time, space, objects, and motion within the video stream, without sending all videos to a central cloud. As the amount of video continues to explode, this native AI architecture is essential for enabling precise, targeted queries rather than a rough workflow that scans every footage.
Merlin: Built-in GenAI assistant with profile support
Merlin, the generative AI agent built into Qualcomm Insight, is the primary operator-system interface. Unlike typical chat-style assistants or simple “AI search bars” added to traditional VMS, Merlin is tightly coupled with the video data model, policies, and profiles defined by the platform.
Traditional systems rely on basic attribute filters layered on top of object detection to support requests such as “Show me the person wearing the red shirt.” Merlin can support more sophisticated profile-based queries, such as “Find Joe in the red shirt.”
Merlin also says this.
Generate incident summaries and reports directly from video and metadata, reducing the overhead of manual reporting.
Supports profile-linked alerts, so instead of bombarding operators with generic motion and object alerts, notifications can be triggered only when a specified entity or behavior matches a specific risk model.
Edge-first architecture for real-time private inference
Qualcomm Insight is built around an edge-first architecture. Inference runs on-device or on a local edge appliance, keeping the raw video in-site and only sending metadata, events, and selectively requested clips to upstream systems. This design reduces latency for real-time decision making, protects privacy, and significantly increases bandwidth and storage efficiency.
Real-time analytics on cameras or appliances without the need for dedicated server racks.
Multi-stream processing for high-density deployments.
Efficiently execute both discriminative models (detection, tracking, and classification) and generative models used in Merlin.
Brownfield and greenfield deployment models
Qualcomm Insight supports both brownfield and greenfield deployments, allowing customers to modernize at their own pace.
Brownfield: For sites with existing low-cost non-AI cameras, the platform uses Edge AI boxes to terminate RTSP/ONVIF streams, run the AI locally, and maintain the existing camera and cable footprint. Video is analyzed at the edge, and only metadata and event clips are propagated to the Qualcomm Insight control plane, enabling “no rip-and-replace” migration to modern VSaaS architectures.
Greenfield: For new builds or complete updates, the platform works with on-device AI cameras that perform inference directly.
Because both models are under a single management and analytics fabric, customers can mix legacy and new sites while maintaining common policies, query semantics, and AI capabilities.
Augentix acquisition expands camera flexibility
This wide selection allows system designers to tailor camera selection to each zone. This means you can have simple low-power devices in low-risk areas, high-performance AI cameras with advanced analytics, and better low-light performance in critical locations, all with the same Qualcomm Insight control plane. The result is a more flexible and cost-effective design without fragmenting the software stack or analytics capabilities.
Multi-vertical and future-proof
Qualcomm Insight is designed as a horizontal platform with vertical focus rather than a series of isolated point solutions. The main examples are:
Standard security and surveillance for buildings and businesses: weapon detection, loitering and tailgating detection, restricted area entry, delivery person and package tracking, and automatic detection and alerts based on configured profiles and policies.
Smart cities and traffic: intersection tracking, accident detection, traffic flow analysis with edge processing on utility poles and roadside units, profile-aware search during investigations.
Retail and Logistics: Loss prevention, queue and holdup analysis, and zone-based monitoring. It has the ability to search for specific individuals or assets based on role, profile, or visual attributes.
Industrial and critical infrastructure: PPE compliance, restricted zone violations, and anomaly detection on industrial lines. Align video closely with OT systems and policies.
AI is native to the platform;
From pre-validated camera SKUs to ready-to-use analytics bundles, we actively work with a growing ecosystem of hardware and software partners to accelerate time to value for our customers. Over the coming weeks, we'll be introducing new features and ecosystem collaborations that expand the capabilities of Qualcomm Insight.
Learn more about Qualcomm Insight Platform
VSaaS
IoT
Edge AI
Generation AI
camera
safety
The opinions expressed in the content published here are the personal opinions of the original author and do not necessarily reflect the opinions of the original author.
About the author
Sri Madhura
Suri Maddhula Vice President, IoT Solutions Product Management;
