Your Vet May Soon Be Asking AI About Your Dog

If your veterinarian pauses during an appointment to check a drug reference, consult a specialist, review a consensus statement or look up an unfamiliar presentation, most pet owners would probably consider that a sign of care rather than uncertainty. Medicine has never required clinicians to hold every fact in their heads at once. It requires them to know what matters, recognize what they do not know, find reliable information and apply it to the living patient in front of them. 

A new tool from Instinct Science could make that familiar act of looking something up feel very different, because instead of searching through a book or database, the veterinarian can simply ask a question.

Instinct Science, a Philadelphia-area veterinary technology company, announced Attending on Sept. 10 as a clinical AI assistant built specifically for veterinarians. The system is designed to accept ordinary-language clinical questions and return answers grounded in a defined collection of veterinary references, including Plumb's Veterinary Drugs, Standards of Care and Clinician's Brief. 

Instinct says every response includes citations that a clinician can inspect, and its product materials describe a knowledge base built from more than 30 years of veterinary content, more than 20,000 primary sources and ongoing review by more than 200 veterinary and pharmacy experts. 

For safety-critical information such as drug dosing, the company says Attending retrieves the information directly from Plumb's rather than asking a generative model to produce a dose from scratch.

That architecture matters because it points toward a more useful way to think about professional AI. Most public anxiety about AI in medicine starts with the image of a general-purpose chatbot confidently making a diagnosis. Attending is being positioned differently. It does not search the open web, and Instinct explicitly says the generated answer is not itself a peer-reviewed publication. The tool is supposed to synthesize and surface vetted material quickly enough to be useful in the moment, while keeping the evidence visible so the veterinarian can decide whether it actually fits the case.

The name is doing some conceptual work too. In a teaching hospital, the attending veterinarian is the clinician responsible for the case, supported by residents, interns, students and other professionals who help gather information and work through possibilities. Instinct founder and CEO Caleb Frankel, an emergency veterinarian, described Attending as an attempt to extend some of that support to clinicians who may be working without a specialist down the hall or a teaching team around them. 

That makes the product less interesting as an artificial veterinarian than as an artificial layer of clinical support, a way to compress the distance between a question and the body of evidence that might help answer it.

For pet owners, that distinction is important. Hearing that a veterinarian is 'asking AI about your dog' sounds very different from hearing that the veterinarian checked a drug handbook or searched the literature, even when the underlying purpose is similar. A conversational interface makes the reference process feel more human because it resembles asking another professional. That resemblance is powerful, and potentially dangerous if the interface earns more trust than the evidence deserves. 

A fluent answer can still be wrong, incomplete or poorly suited to an individual animal. Citations do not eliminate that problem, but they make the answer inspectable in a way that a confident, source-free chatbot response is not.

Instinct is also drawing a firm line around responsibility. Its product page says diagnosis, prescribing, treatment decisions and client communication remain the veterinarian's job. That is not just a legal distinction; it is the center of the model being proposed. The system can help with recall, comparison and research, but the clinician still has to interpret symptoms, examine the animal, understand the medical history, weigh uncertainty and account for the practical realities of care. A recommendation that makes sense in a journal article may not make sense for a specific patient, a specific family or the resources available in a specific hospital. Clinical judgment lives in that gap between information and action.

Attending is not yet a routine live-care tool. Instinct says the system is in a research preview with hundreds of veterinarians and has been trialed by specialists, emergency clinicians and general practitioners, including at New York's Schwarzman Animal Medical Center. The company's current terms of use still classify Attending as a pre-release beta and prohibit its use in a live clinical environment or in connection with the actual care of a patient. That boundary is worth keeping in view, because it separates what the product is demonstrating now from what Instinct says it intends to build next. The company plans to offer Attending as a standalone product and eventually embed it directly into Instinct EMR and ScribbleVet workflows.

That eventual integration may be the more consequential part of the story. Once a clinical reference assistant sits inside the same system where a veterinarian reviews records, writes notes and manages a case, 'looking something up' stops being a separate task. The reference layer becomes part of the workflow itself. Instead of remembering which database to open, translating a clinical question into search terms, scanning multiple pages and then carrying the answer back into the record, the veterinarian could ask the question where the work is already happening. The gain is not simply speed. It changes where professional knowledge appears and how frictionless it becomes to consult it.

There is a larger cultural shift hiding inside that convenience. We tend to treat expertise as though it lives entirely inside the expert: the doctor knows, the lawyer knows, the mechanic knows. In practice, expertise has always included tools, references, colleagues, institutional memory and the judgment to use them correctly. 

AI makes that scaffolding more visible because the tool talks back. It can feel as though the knowledge has moved out of the professional and into the machine, when a better description may be that the professional has acquired a faster interface to a body of knowledge they were already expected to consult.

That does not make the interface neutral. The easier it becomes to ask, the easier it may become to accept. Good clinical AI will therefore be judged less by how convincingly it sounds like a specialist and more by whether it helps a specialist, emergency doctor or general practitioner see the evidence, notice uncertainty and retain control of the decision. Instinct's emphasis on source visibility, curated veterinary material and direct retrieval for dosing is clearly designed around that problem. 

The real test will be whether those safeguards remain useful when the product moves from a research preview into the rushed, interrupted, emotionally complicated reality of everyday veterinary medicine.

So yes, your veterinarian may soon ask AI about your dog. That sentence will probably sound less strange over time. The meaningful question is what kind of AI they are asking, what evidence sits underneath the answer, whether they can inspect it, and who remains responsible for deciding what happens next. If systems like Attending succeed, the future of veterinary AI may not look like a robot replacing the person in the exam room.

It may look much more ordinary: a veterinarian doing what good clinicians have always done, checking the evidence before making the call, with a new kind of reference tool sitting beside them.

SOURCE NOTES

• Instinct Science, Sept. 10, 2026: 'Instinct Science Announces Attending, a First-of-Its-Kind Clinical AI Assistant Built Exclusively for Veterinarians'
• Attending by Instinct product page
• Caleb Frankel, VMD, Sept. 9, 2026: 'Introducing Attending: Building Support Around Veterinarians'
• Instinct Terms of Use, last revised Sept. 1, 2026