The Problem: Nutritionists at Humess (a genetic analysis and health optimization company) were spending over 40 hours a week manually cross-referencing complex genetic test results, dietary logs, and medical histories just to answer routine client questions. Generating scientifically backed, personalized recommendations was a slow, unscalable process. It created a massive bottleneck in patient management and strictly limited the number of clients a single medical specialist could handle.

The Tech Stack:

• Core Orchestration: n8n

• AI Engine: OpenAI ChatGPT (RAG Architecture)

• Database / Knowledge Base: Notion

• Client Interface: Telegram Bot API

The Zore AI Fix: Instead of forcing medical staff to dig through scattered PDFs and databases, we built a centralized AI Nutritionist Agent using a Retrieval-Augmented Generation (RAG) pipeline. All client data—including genetic markers, meal plans, and medical histories—is securely structured inside Notion. When a client asks a question via the Telegram Bot, the n8n orchestrator instantly fetches that specific patient's medical context from Notion and feeds it to ChatGPT. The AI generates a hyper-personalized, clinically accurate response based on the client's unique DNA and diet. The Humess specialist simply reviews the AI-generated draft (human-in-the-loop) and approves it for sending.

Data Processed Automatically: Genetic markers, DNA predispositions, dietary histories, medical questionnaires, and real-time client queries.

The Hard Numbers:

• 40+ hours per week saved on manual data retrieval and patient research.

• Response time slashed from several hours to under 30 seconds for personalized nutritional advice.

• 3x increase in patient capacity per nutritionist, without sacrificing the quality of personalization.

Hot take: Paying highly qualified medical professionals to manually compile data from PDFs and databases is a waste of their expertise. AI should gather the context; the specialist should make the final decision.