What counts as a source
Textbooks. Peer-reviewed journals. University veterinary references. The guidelines vets follow. FDA-approved drug labels.
Nothing from a search engine, a blog, or a forum.
Not all answers are built the same.
Everything BarkLens tells you about your dog comes from one of two places: your dog’s own records, or a published veterinary source that a licensed veterinarian approved. Not the open internet, and not whatever a model happens to remember.
Everything BarkLens tells you comes from your dog’s own records, or a veterinary source a licensed vet approved.
→ Your dog isn’t the average dog
→ No message boards. Ever.
→ A vet decided what counts
→ You don’t have to trust us
Merck Veterinary Manual · Cornell University eClinPath · Animal Health Diagnostic Center · AAHA practice guidelines · ACVIM consensus statements · ACVS · IRIS · FDA-approved drug labeling
Merck · Cornell eClinPath · Animal Health Diagnostic Center · AAHA · ACVIM · ACVS · IRIS · FDA drug labels
Textbooks. Peer-reviewed journals. University veterinary references. The guidelines vets follow. FDA-approved drug labels.
Nothing from a search engine, a blog, or a forum.
A licensed veterinarian reviews every source before it’s used.
They’re not reviewing your dog’s results. Those belong to your own vet, who has examined your dog. What gets reviewed here is the material behind the answers.
Not your dog’s results. Those belong to your own vet.
Every claim shows the passage it came from, word for word, and where it was published.
If a source is unclear or contradicts itself, the claim doesn’t get used.
It won’t diagnose. It won’t tell you what’s wrong with your dog, and it won’t tell you nothing is.
It explains, organizes, and gives you the questions worth asking. Your vet makes the calls.
BarkLens uses AI and can be wrong. That’s why every claim points to a source you can check.
In a study of 636 citations from AI-generated papers, 55% of the works cited by GPT-3.5 were fabricated (Scientific Reports). In another, eight AI search tools were asked to name the source of a direct quote across 1,600 queries and got it wrong more than 60% of the time (Columbia Journalism Review).
Newer models do better. We still think you should be able to check.
In one study, 55% of the works cited by GPT-3.5 were fabricated (Scientific Reports). Newer models do better. We still think you should be able to check.
BarkLens doesn’t diagnose. Your vet does.