Free for verified vets, techs & students

A colleague you canaudit.

Ask it like the colleague across the hall — then watch it grade every citation and hold back any dose it can't source, before you act.

How it works

Your clinical copilot,from first question to finished note.

  1. 1Ask
    Holland Lop, anorexic 24hr, hunched…

    Type your clinical question or paste a patient case.

  2. 2Verify
    Verdict
    DrugDose

    Read the cited verdict, recommendations, and dose math.

  3. 3Document
    SOAP DraftCited
    S
    O
    A
    P

    One click generates the SOAP note or client handout, citations intact.

What happens behind the answer

One answer. Specialists behind it.

Separate agents reason through the case, carry the patient's history, verify claims, gate sources, and guard dose math before the answer reaches you.

  • Reasons the case
  • Remembers the patient
  • Re-checks every claim
  • Gates the sources
  • Guards the dose
  • + more
Trust you can watch

It doesn't just answer.A second agent re-checks every claim.

After every answer, a second agent re-checks it section by section against the literature — hedging weak claims, deleting any number it can't source, re-checking every citation — then shows you exactly what it changed. Nothing is edited silently.

GI stasis · Holland LopReviewingv0v1

Reviewing every section…

Uncited number
GI stasis recurs in ~40% of rabbits after a first episode.

No source cleared the bar — the figure is removed, not guessed.

Citation check
Motility returns within 24–48 h of prokineticsK5Directly supports

Source is topical, not direct support — demoted, never hidden.

Weak claim
Cisapride shortens time to first fecal outputK3

Backed only by a single case report — hedged, and the source labelled.

Every version is kept — nothing is edited silently; every change is yours to inspect.

Every section, every answer — re-checked against the literature.

The memory agent

It remembers the patient.Not just the chat.

Most AI forgets the moment you close the tab. Voyage runs a separate memory agent that keeps a structured record of every patient — prior visits, labs, the meds you tried and how they went — decides what's worth keeping, and pulls the relevant history into the next case. It's tuned to your clinic, and it deepens every visit.

Bella · Golden Retriever, 4yBuildingpatient record
MarAtopic dermatitis
started Apoquel
JunSeasonal flare
Cytopoint — good response
AugOtitis externa
Malassezia on cytology
TodayRecurrent pruritus, ears again

A structured record for every patient — you never brief it from zero; every visit adds to it.

  • Keeps what matters

    Decides what's clinically worth retaining — not a transcript dump.

  • Recalls per case

    Surfaces the relevant history for the patient in front of you.

  • Structured, per patient

    A real record it maintains, not a fuzzy summary.

  • Learns your clinic

    Tuned to your practice's patterns — not a generic average.

Grounded in the veterinary canon

The veterinary literature —indexed, cited, and growing weekly.

Every Voyage answer draws from a curated corpus of textbooks, peer-reviewed journals, formularies, and consensus statements — 0+ veterinary-literature sources, grounded in the literature, not the open web.

"If it's in Plumb's, Ettinger, JVIM, or an ACVIM consensus statement — it's in the corpus we cite from."

Standard textbooks

  • Plumb's Veterinary Drug Handbook
  • Ettinger — Veterinary Internal Medicine
  • Nelson & Couto — Small Animal Internal Medicine
  • Bonagura — Kirk's Current Veterinary Therapy
  • Quesenberry & Carpenter — Ferrets, Rabbits & Rodents
  • Mader — Reptile and Amphibian Medicine
  • Slatter — Textbook of Small Animal Surgery
  • Tilley — 5-Minute Veterinary Consult
  • BSAVA Small Animal Formulary
  • Carpenter — Exotic Animal Formulary

+ 50+ canonical references

Peer-reviewed journals

  • JVIM — J. of Veterinary Internal Medicine
  • JAVMA — J. of the American Vet. Medical Assn.
  • Veterinary Surgery
  • Journal of Feline Medicine and Surgery
  • Journal of Small Animal Practice (JSAP)
  • Journal of Exotic Pet Medicine
  • Journal of Avian Medicine and Surgery
  • AJVR — American Journal of Veterinary Research
  • Veterinary Clinics of North America
  • Veterinary Radiology & Ultrasound
  • Veterinary Anaesthesia and Analgesia

+ 200+ peer-reviewed sources indexed via PubMed

Consensus & clinical guidelines

  • ACVIM Consensus Statements
  • IRIS CKD Staging & Management Guidelines
  • WSAVA Vaccination, Dental & Nutrition
  • AAFP Feline Practice Guidelines
  • AAHA Anesthesia, Pain & Vaccination
  • ISFM Feline-Friendly & Senior Cat Guidelines
  • IVAPM Pain Management Guidelines
  • AEMV Exotic Mammal Practice Guidelines
  • ECVIM-CA Consensus Statements

+ active tracking across specialty colleges

+ Expanded weekly by our clinical content team. New sources added as published in the literature. View the full source manifest →

Open veterinary AI benchmarks

Measured in the open. Built to be inspected.

Two reproducible system evaluations test whether clinical claims carry authoritative support — and whether memory stays attached to the right pet through change and interference.

50
evidence cases
24
memory tasks
100%
open score inputs
Track 01VetEvidenceBench 2.550 paired cases

More clinical claims you can actually audit.

Voyage delivered substantially more risk-weighted, citation-backed content across the same 50 cases. The coverage advantage survives the prespecified competitor-favourable missing-evidence bound.

Citation precision
67.3%
High-risk authoritative
45.8%
Claim coverage
29.0%
This supports a comparative coverage claim, not “more accurate citations.” Absolute judge-derived values remain provisional until blinded veterinary calibration is complete.
Read the full scorecard

Risk-weighted coverage · higher is better

49.3%

Voyage v3.2

Voyage v3.249.3%
GPT-5.6 Sol7.9%
Claude Opus 52.5%

10,000-iteration case bootstrap

Track 02PetMemoryBench 2.024 tasks × 3 recalls

The right memory. For the right pet.

A multi-pet, high-interference workload probes current state, changed facts, cross-pet isolation, and compound continuity.

Voyage72/72100% answer accuracy
ChatGPT Free63/7287.5% answer accuracy

This small synthetic cohort is descriptive, not a significance claim. Voyage’s selected-pet routing favours Voyage; this dated product comparison does not represent every model or memory setting.

Read the full scorecard
0112/12Current state
0212/12History & change
0312/12Cross-pet isolation
0412/12Compound continuity

Inspect every frozen prompt, adjudication, schema, and dependency-free scorer — then run your own system against the same contracts.

Try a real case

Scenarios from the trenches.

Before you apply

The questions worth asking.

Yes. Access is free for licensed veterinary professionals — veterinarians, credentialed technicians, and veterinary students. There is no card and no paid tier on the clinical desk.

Veterinarians (DVM, VMD, BVSc, BVMS, MRCVS), credentialed technicians — LVT, RVT or CVT, whichever title your state uses — and veterinary and vet-tech students. You create an account, then submit your credential: licence number and jurisdiction for veterinarians and technicians, or a school email for students. The application is reviewed before the clinical desk opens.

Yes. Clinical claims carry a citation to the veterinary source behind them, so you can check the reasoning rather than trust an unattributed answer. Sources are listed with each answer and the reference base is published at thevoyage.ai/forvets/sources.

No. Voyage is a reference and reasoning tool that surfaces cited evidence to inform your decision. It does not examine the patient, does not establish a veterinary-client-patient relationship, and does not replace your clinical judgment. Confirm dosing and treatment against current formularies before you act.

Dogs and cats, plus exotic companion species — rabbits, guinea pigs, ferrets, rodents, birds, and reptiles. Exotic coverage is a deliberate focus rather than dog guidance reshaped to fit another species.

Every dose carries its source and an always-on instruction to verify against your current formulary — that line is the floor, not a formality. A separate guard catches the transcription failures a free-text weight invites: a cat on a dog-intended dose, or a weight an order of magnitude out. It stays deliberately silent for equine, production, and exotic species, where there is no defensible weight band to check against — so the absence of a warning is never a validation, and the formulary check is still yours to make.

Yes. From any answer you can generate a SOAP note or a plain-language client handout in one step, with the citations from the underlying answer carried through.

A general-purpose assistant will answer a dosing question fluently whether or not a source supports it, will do it for a sulcata as readily as for a Labrador, and gives you no way to check either answer. Voyage is built so the evidence is inspectable: claims are cited, sources are named, the reference base is published rather than implied, and exotic species are covered on purpose rather than by analogy to the dog.

Yes. VetEvidenceBench 2.5 publishes 50 paired veterinary cases, and PetMemoryBench 2.0 publishes 24 multi-pet memory tasks with three recalls each. The frozen prompts, dated product captures, adjudications, schemas, score inputs, methods, and offline scorers are public. These are system evaluations: each result describes the tested product, model, tools, memory configuration, and run protocol together — not an underlying language model in isolation.

Case details you enter are used to answer your question and to keep context across a case — not sold. Voyage is a working aid, not the medical record of record; your practice management system remains the custodian of the patient record.

For working clinicians

Ready to see every claim traced to its source?

Create your account

Free for verified vets, techs & students · No card, no waitlist.