Veterinary AI as a research frontier
Veeeet is a deployed pet health platform. Behind it is a research agenda at the intersection of veterinary medicine, multimodal health data, and adaptive AI systems.
What the technology actually does
The core of Veeeet is not an LLM. It’s a system that turns unstructured, subjective owner observations into longitudinal, structured health data — for animals that cannot describe their own symptoms.
This requires combining:
- Computerized Adaptive Testing (CAT) — questionnaire logic that adapts based on prior inputs
- AI Vision — automated extraction and interpretation of veterinary documents
- Rule-based veterinary logic — risk flags and clinical thresholds developed with practicing vets
- LLM layer — for personalized interpretation of already-structured data
The result is a continuously updated digital twin of each individual pet. Remove the LLM — the system still works. That distinction matters.
Four open research questions
These are the problems where we need research partners.
1. Multimodal data fusion for longitudinal health profiles How do you combine heterogeneous inputs — clinical records, behavioral observations, owner-reported data, document scans — into a coherent, dynamically updated model of an individual animal’s health state?
2. Early disease signal detection from behavioral patterns Pets can’t report symptoms. But longitudinal behavioral data holds early signals — subtle changes in activity, appetite, social interaction. How do you reliably extract clinically meaningful patterns from noisy, owner-mediated data over time?
3. Explainability in veterinary AI recommendations When an AI system recommends «see a vet urgently», both the owner and the clinician need to understand why. How do you build meaningful explainability into a hybrid rule-based and LLM system — in a context where clinical trust is non-negotiable?
4. Multi-agent consistency under conflicting signals Our architecture uses specialized agents — symptom analysis, nutrition, medical history — feeding a judge model. When agents disagree, how do you ensure the final recommendation is safe, consistent, and verifiable?
What we bring to a collaboration
A deployed product with 1,000+ real users on iOS and Android — providing a real-world research context, not synthetic data.
Multimodal longitudinal data: structured questionnaire responses, veterinary document scans, behavioral observations over time, breed-specific health flags.
A clinical foundation: our veterinary advisory team brings 40+ combined years of practice. Questionnaire logic and risk flags are developed with practicing veterinarians — not derived from general medical datasets.
Openness to joint publication, Horizon Europe and national R&D grant applications, and shared IP arrangements where appropriate.
External validation
Acelero acceleration program — EU-funded (NextGenerationEU). Innovation confirmed across technology patentability, AI methodology, and scalability.
Global Startup Awards — Regional Finalist, Southern Europe, 11th Edition.
WebSummit 2024 — HealthTech Revolution track.
F6S Pet Tech Spain — Top 3 (November 2025), Top 10 consistently since launch.
Let's talk
If you work on adaptive AI systems, health data fusion, clinical explainability, or multimodal modeling — and any of these questions intersect with your research — we’d like to connect.
We can offer a 20-minute call, a technical summary, or a detailed concept note depending on what’s useful.
Whatsapp: +34 625 839 238
Linkedin: https://www.linkedin.com/in/vadimgusakov/