Written by Dr. Hannah Weitzenfeld
If you’ve ever spent an hour after your shift writing adoption summaries, drafting foster instructions for ringworm kittens, or answering emails about treatment plans, you’ve probably wished for an extra set of hands — ideally ones that don’t need to be scrubbed in. AI can’t examine a patient, interpret shelter dynamics, or make welfare decisions, but it can help with the administrative tasks that quietly eat a shelter veterinarian’s day like a cat with unsupervised access to Churu.
Medical documentation. A free account with ChatGPT, Claude, or Gemini is enough to start. Give it the clinical shorthand from the chart (“3mo DSH, grade II/VI left systolic murmur, asymptomatic, recheck in 4-6 weeks”), and it returns a plain-language paragraph for the adopter — explaining the finding, recommending follow-up, and reassuring them it’s manageable. You’re not asking it to interpret the murmur, just to translate your interpretation. You still review and edit before anything touches the medical record.
Outbreak management. During kitten season, when your foster coordinator asks for a handout for fosters with recovering panleukopenia kittens, paste your existing protocol into Claude or ChatGPT (both support file uploads, as does Microsoft Copilot if your shelter runs Microsoft 365) and ask it to adapt it: “turn this isolation protocol into a one-page handout for foster volunteers, 6th-grade reading level, with a checklist.” You’re reformatting a protocol your medical team already approved, not inventing a new one — which frees up the afternoon for evaluating cats instead of fighting Word’s bullet points.
Data analysis. Shelter software (Shelterluv, PetPoint, Chameleon) exports intake and medical data as CSVs, which are their own special form of suffering to read manually. Claude and ChatGPT’s data analysis features can take that export and answer plain questions: “What’s our average URI length of stay since the foster program started?” Always verify the actual numbers against the source data before presenting them anywhere — AI spots patterns well but isn’t reliable at arithmetic it doesn’t show its work for.
Communication. Shelter vets talk to adopters, foster caregivers, rescue partners, and leadership — sometimes all in the same hour, usually about the same escaped rabbit. A chatbot is genuinely useful for first drafts: a delayed-surgery email, a board report, a home-care handout. Pasting in a couple of your shelter’s past emails as a style reference (“write this in the tone of the email below”) cuts editing time considerably. You still send it under your own name, so you still own every word.
Of course, AI has limits. It can produce inaccurate information, cite nonexistent references, or confidently recommend the wrong approach — confidently being the key word, since it rarely flags its own uncertainty. It should never make diagnoses, set treatment plans, calculate drug dosages without verification, or make outcome decisions. Clinical judgment and shelter experience remain irreplaceable.
It’s also worth checking whether your shelter software vendor offers an AI feature with an actual data agreement before pasting real adopter names or case details into a free consumer chatbot.
For shelter veterinarians, AI’s value isn’t doing new things, but doing familiar things faster: an adoption summary, foster instructions, a kitten-season data summary. It hands back something often in short supply — time, and maybe the will to answer one more email about whether glitter is toxic to dogs (it’s not great, but the dog will survive; so will you).
The future of shelter medicine still depends on skilled professionals balancing medicine, welfare, population management, and public expectations. AI is just another tool in the toolbox — one that, used thoughtfully, means less time at the keyboard and more time where it matters: with the animals.