Narwal Flow 2 Debuts High-Suction Robovac With Item-Detection AI

Narwal used its CES 2026 presentation to position the Flow 2 as a major step forward in home robotics, blending what the company bills as industry-leading suction with a camera-driven artificial intelligence system designed to find small, valuable items. The unit is advertised with 30,000 Pa of suction, LiDAR-based mapping and navigation, and two HD RGB cameras whose combined field of view Narwal describes as covering roughly 136 degrees.

Engineered as a successor to the company’s earlier Flow mop and an evolution of the Freo series, the Flow 2 ships with a refreshed physical design and an upgraded sensing stack. Narwal has characterized the camera and AI system as enabling "unlimited" object recognition, an expansive claim intended to set the device apart from rivals that support smaller, preconfigured object libraries.

A central feature demonstrated at CES was the Flow 2’s lost-item detection workflow. In demonstrations, the robot approaches a detected object and stops about 5 centimeters away to avoid sucking it up. It then sends a photograph and the item’s mapped location to the user via a push notification in the Narwal app. Narwal and early coverage highlight the system’s ability to detect items homeowners commonly lose on floors, including jewelry, wallets, phones and keys, and the same sensors can also locate pets for remote monitoring.

The device does not autonomously pick up or carry items. Demonstrations and early reporting make clear that the Flow 2’s so-called retrieval capability is primarily detection and reporting: the robot tags and maps objects and notifies users rather than physically retrieving them. Users also cannot currently dispatch the Flow 2 on a directed "search mission" to hunt for a specific item on command. That limitation frames the innovation as an advanced awareness tool rather than a robotic valet.

Narwal has added lifestyle-focused cleaning modes that reflect the company’s consumer positioning. The Flow 2 introduces Pet Care Mode and Baby Care Mode, with marketing language calling the device "the most advanced robotic mop ever" and saying it is "prepped" for homes with babies. Baby Care Mode is described as lowering noise as the robot approaches a crib and sending reminders about misplaced toys, while Pet Care Mode supports locating animals when owners are away and tailoring cleaning behavior around pet areas.

The company also emphasized navigation resilience: LiDAR mapping, AI-driven route planning, and on-device updates designed to recalibrate paths when obstacles appear. As with previous models, Narwal and observers advise that sensor performance improves in less cluttered environments and with routine sensor maintenance.

Narwal’s claims at CES represent product positioning and early demonstrations rather than results from independent testing. Availability, pricing and long-term reliability were not disclosed in the presentation. Beyond convenience, the Flow 2 raises ethical and privacy considerations tied to persistent imaging and mapping inside private homes, along with questions about the accuracy of automated object recognition and the risks of false positives or overlooked valuables. For many households, however, the combination of stronger suction and smarter sensing could meaningfully reduce the everyday frustration of searching for lost items and managing pet- and baby-focused cleaning needs.