What is it?
Less guesswork. More horsepower.
Smarter training, clearer feedback, and every session moving forward.
Carrots is the future. Don’t wait for someone else to show you how it works.
In a nutshell - You give us your session data. We run it through our system. You get clear feedback on fitness, recovery, and what to tell the owners — without needing to crunch the numbers yourself.
Behind every recommendation is a tightly integrated LLM-powered RAG pipeline — a modular processing stack designed to retrieve, interpret, and respond with scientific precision. We’ve engineered Carrots to act as a real-time equine performance engine, ingesting multimodal training data and outputting structured, context-aware insight in seconds.
At the core is a fine-tuned Large Language Model layered with Retrieval-Augmented Generation (RAG) — meaning every output is grounded in source-backed, citation-locked material drawn from a purpose-built vector database of over 50+ peer-reviewed equine physiology studies, lactate profiling tables, VO₂max thresholds, and training periodisation schemas. No hallucination. Just relevance.
Your training data — whether streamed from E-Trakka, Polar, or structured manual entry — is parsed, tagged, and embedded via Carrots’ internal preprocessing layer. This allows dynamic query routing and longitudinal comparison across individual horses or entire stables, stored securely in a high-availability, low-latency Supabase instance optimised for structured physiological data access.
Every insight is backed by evidence. Every flag is explainable. And every trend is machine-readable, trainer-actionable.