Nutrition Agent is a chat-based AI assistant designed to provide personalized, adaptive, and intelligent dietary guidance. Built using IBM’s Granite-3-3-8b-instruct model, LangGraph, and ReAct architecture, the agent delivers real-time meal planning, food alternatives, and fitness-aligned nutrition advice—all through natural language conversations.
- 🍱 Personalized Meal Plans based on user health goals and preferences
- 🔄 Smart Food Swaps with context-based reasoning
- 🗣️ Natural Language Interaction – just chat like you're talking to a real dietician
- 🎯 Goal-Aware Guidance – supports fat loss, muscle gain, and healthy living
- 🧾 Explanation Engine – every suggestion is backed by logic
- 📊 Scalable & Adaptive – no human appointments needed
| Technology | Description |
|---|---|
| 🧠 IBM Granite Model | Instruction-tuned LLM (granite-3-3-8b-instruct) |
| 🧪 LangGraph | Workflow orchestration and agent logic chaining |
| 🧠 ReAct Architecture | Multi-step reasoning and acting |
| ☁️ IBM Cloud Lite | Model hosting and deployment |
| 💬 NLP | For understanding user input and generating responses |
- IBM Watsonx AI Studio
- IBM Watsonx AI Runtime
- IBM Granite Foundation Models
- IBM Cloud Agent Lab
-
User opens the chat interface and types a query like:
“Suggest a vegetarian meal plan for weight loss.”
-
The agent extracts context (diet type, goal) and generates a day-wise meal plan.
-
The user can follow up with:
“Swap rice with a healthier option.”
And the agent suggests:
“Quinoa is a healthier choice—it’s high in protein and fiber.”
- 🌐 Multilingual & voice support
- 🤝 Integration with dieticians/fitness trackers
- 🔍 Real-time nutrient gap detection
- 🧾 AI-generated nutrition reports
- 🖥️ Edge deployment for offline mode
- IBM Cloud: IBM Cloud Lite
- IBM Watsonx.ai: IBM Watsonx.ai
- IBM SkillsBuild: IBM SkillsBuild
Mahaprasad Sahoo
MCA, Department of Engineering
GITA AUTONOMOUS COLLEGE, BHUBANESWAR
Github Link : https://github.com/mahaprasadsahoo12
➡️ https://github.com/mahaprasadsahoo12/Nutrition_Agent
All Agent previews/Screenshots related to this project is available under the Agent_Preview/ folder.
All IBM Certifications related to this project are available under the Certificates/ folder.
Feel free to ⭐ star this repo if you found it helpful.






