Hoomanely's EverBowl: A Smart Dog Feeder That Watches for Signs of Illness
Palo Alto startup Hoomanely has built a sensor-packed feeding station for dogs that tracks food and water intake, eating speed, chewing and swallowing sounds, facial temperature, and mouth movement patterns, then feeds that data through an AI system to build a health baseline for each dog and flag changes that could signal illness. The company has shipped roughly 50 units of its EverBowl device so far, offered as a hardware-plus-app combination for $29 per month. CEO and co-founder Sai Supriya Sharath outlined the product in an interview with TechCrunch published August 27, 2026. TechCrunch
The EverBowl processes data on the device itself rather than in the cloud, using what Hoomanely calls its VBus architecture — a combination of sensor fusion (merging data from multiple sensors into a single picture) and edge AI (machine learning that runs locally on the device, avoiding the latency and privacy trade-offs of sending data to a remote server). Raw sensor signals flow into the company's BioSense AI Engine, which Hoomanely describes as a deep machine learning and large language model-powered system that turns the data into preventive-care insights. Those insights reach owners through the EverWiz app, a patented, canine-focused health assistant. The wider ecosystem has three layers: the Physical Intelligence devices (Hoomanely's term for its sensor-equipped hardware), the BioSense AI Engine that processes the data, and a community feature called the Pet-Parent Exposome for sharing care experiences.
The platform was developed over an 18-month beta testing phase during which Hoomanely collected roughly 5 million data points across more than 80 dogs. During that period, the system flagged changes in one dog's eating patterns that turned out to be connected to a chipped tooth becoming infected. In a separate beta case, the app showed a sustained deviation from a dog's eating and temperature baselines, prompting the owner to visit a vet; the dog was diagnosed with tick fever. Both cases were reported by TechCrunch on August 27, 2026. TechCrunch
Hoomanely is straightforward about the current limits of its clinical evidence. The company says it does not yet have independent sensitivity (how well the system correctly detects real problems), specificity (how well it avoids false alarms), or false-positive rates for clinical events, because producing those figures would require a formal study comparing its alerts against actual veterinary diagnoses. The company states that the platform is not a replacement for veterinary diagnosis but rather a tool to help owners and vets identify potential issues earlier. To address that evidence gap, Hoomanely is using its accumulated data for formal veterinary studies measuring sensitivity, specificity, positive predictive value, and false-positive rates for each alert category, with results to be reported by use case. TechCrunch
The hardware roadmap extends beyond the bowl. Hoomanely plans EverSense, a wearable that would capture movement and rest data, and EverHub, a hub device designed to accept inputs from third-party hardware such as smart collars, feeders, and home environmental sensors. The company positions this as a step toward becoming an animal health data company rather than a single-device vendor. The founding team includes COO and co-founder Harshal Hinger and CTO and co-founder Vipin Ravindran. Hoomanely's website lists a dog named Champ as Chief Inspiration Officer.
The broader context here is the trajectory of continuous health monitoring. We have watched this approach move from clinical settings into wrist-worn devices and now into a bowl-mounted form factor, and the pattern is consistent: gather dense longitudinal data, build an individualized baseline, and alert on deviation. The veterinary angle is particularly interesting because dogs cannot describe their symptoms, making passive, always-on sensing comparatively more valuable than it is in human health, where patients can articulate what they feel. A feeding station that captures the acoustic profile of chewing, facial heat readings, and consumption speed is a rich multi-modal sensor package for a single point in the home.
In my view, the gap between compelling beta anecdotes and clinically validated alerting is where most continuous-monitoring products stall. Two cases from an 80-dog beta are illustrative, not statistically meaningful. Hoomanely's own acknowledgment that it lacks sensitivity, specificity, and false-positive data is the honest framing, and the formal veterinary studies it has initiated are the right next step. Until those results are available, the $29-per-month subscription buys a sophisticated sensing platform with plausible but unvalidated clinical utility. The 50 devices shipped represent an early stage, and the company's long-term viability as a health data company will depend on whether the BioSense AI Engine's alerts hold up under the kind of prospective, diagnosis-anchored scrutiny that Hoomanely says it is now pursuing.
