What pet mood analyzers actually measure
The tools sold as pet mood analyzers are doing at least four unrelated jobs. Most of the disappointment in the reviews comes from buying one kind while expecting another.
Search for a pet mood analyzer and the results come back as one undifferentiated list. They are not one thing. There are at least four distinct products in that list, they read different signals, and a good share of the anger in the reviews is people who bought one kind expecting another.
Four products under one name
Audio translators. You record a meow or a bark and get a sentence back — "I'm hungry", "I want attention". This is the largest group by installs and by review count. It is also the one with the worst reviews, and we have argued elsewhere that the problem is structural rather than a matter of accuracy: a meow is a request aimed at you, not a report of an internal state.
Body-language readers. You upload a photo or a short clip and get a reading of posture — ears, eyes, tail, where the weight sits. Hachimo is one of these. So are several others, and they differ from each other more than the category name suggests.
Activity trackers. A collar sensor logging movement, sleep and scratching across weeks. These do not read mood in a moment at all. They read change over time, which is a narrower question and in some ways a more answerable one.
Novelty filters. A caption over your pet's face, played for laughs. They are not claiming accuracy and it is unfair to judge them as if they were. The trouble is that search puts them next to the ones that do claim it.
Why the category matters more than the accuracy figure
Several of these tools advertise a percentage. The number is close to meaningless on its own, because none of them tell you what it was measured against.
Accuracy against what? There is no ground truth for how a cat feels. You cannot ask it. The usual method is agreement with human raters, which measures whether the model matches what people think they see — a real thing to measure, but not the same thing as whether the cat is content.
The category question is answerable and the accuracy question mostly is not. If a tool is reading the wrong signal, no accuracy figure rescues it.
Five questions before you trust one
Each of these is answerable in about a minute on the product's own page.
Does it tell you which cues it used? A tool that returns "Happy, 87%" and stops has given you a verdict you cannot check. One that says which ear position and which tail height it saw has given you something you can look at yourself and disagree with. The second kind is more useful even when it is wrong, because you find out that it is wrong.
Does it ever refuse? Point a body-language reader at a photo of a dark room and see what happens. Whatever it does with an unreadable image is what it will do with your real ones. The willingness to say "not enough to go on" is the closest thing to an honesty signal you can test in thirty seconds.
Does it separate pain from mood? A hunched back, front paws forward rather than tucked, head carried below the shoulder line — that is a pain posture. It should stop the reading and point at a vet, not come back scored as sadness. Anything that scores it as an emotion is doing something worse than being inaccurate.
Does it keep a history? One reading tells you very little. Mood is legible mainly as change, and change needs something to change from. A tool with no record cannot answer the only question that reliably matters, which is whether this is different from how your animal usually looks.
What happens to the photo? Read the privacy policy for whether images are stored, for how long, and whether they train anything. This one is not about quality, and it is the one people check least.
Where we sit in this
We make Hachimo, so treat what follows as an interested party describing its own design choices rather than a verdict.
It is a body-language reader: a photo or a short video in, a mood score with the specific cues behind it out. It does not listen to anything. It declines images it cannot read rather than inventing a mood for them, and it flags pain postures as a vet question instead of scoring them.
Those are answers to three of the five questions above. The other two — history and what happens to your photo — are worth checking for yourself, on us and on anyone else, rather than taking from a page we wrote.
Made with 💛 and many head scratches. Hachimo is for fun and curiosity — it's not veterinary advice. If your pet seems unwell, please talk to a real-life vet. 🩺