A super lightweight, local-first chatbot designed for desktop/laptop use — ideal for minimal setups, offline usage, and quick experimentation.
🛠️ This is my first chatbot project, later refined and optimized before being shared publicly for others to explore, learn from, and build upon.
This chatbot is intentionally designed to run locally using Ollama instead of cloud-based AI APIs.
Why local-first?
- 💸 No API or usage costs
- 🔒 Improved data privacy
- 🔌 Fully offline-capable
- 🧠 Full control over the AI model and prompts
- 🧪 Ideal for experimentation and learning system architecture
Due to hardware and runtime requirements, this project is not deployed on free cloud hosting platforms and is meant to be run on personal machines.
- 💬 Local LLM chatbot powered by Ollama
- 🧠 Custom persistent memory using
memory.json - 📝 Chat and journal history stored in
chat_history.json - ⚡ Lightweight and clean architecture
- 🌐 Accessible via browser on the same local machine or local network
- 📦 Uses models pulled directly from Ollama (no LM Studio required)
- 🖥️ Optimized for laptops and desktops
- 🔌 Offline-capable
- 🔧 Built for simplicity and future extensibility
Developed and tested on:
- Intel Core i5-13420H
- RTX 4050 Laptop GPU
- 16GB RAM
- Windows 11
Minimum recommended:
- Quad-core CPU
- 8–16GB RAM
- Optional discrete GPU
- Smaller quantized models (e.g. Phi-2) work well on modest hardware
app.py → Main Flask application brain.py → Core chatbot logic
functions/ ├─ history_func.py → Conversation history handling ├─ journal_func.py → Journaling logic ├─ memory_func.py → Persistent memory system ├─ model_runner.py → Ollama model interaction └─ prompt.py → Prompt construction
templates/ ├─ index.html → Main chat UI └─ history.html → History & journal viewer
static/ ├─ style.css → Styling └─ script.js → Frontend logic
memory.json → Persistent chatbot memory chat_history.json → Stored chats & journals demo.gif → UI preview
-
Install Ollama
https://ollama.com -
Pull a model (example: Phi-2) ollama pull phi:2
-
Run the model ollama run phi:2
Phi-2 is used for its balance of speed and performance.
You may also use models like mistral or llama3.
-
Clone the repository -git clone https://github.com/Pranziss/Simple--Lightweight-Chatbot.git -cd Simple--Lightweight-Chatbot
-
(Optional) Create a virtual environment -python -m venv venv venv\Scripts\activate # Windows source venv/bin/activate # Mac/Linux
-
Install dependencies -pip install flask
-
Start Ollama and run the app ollama run phi:2 python app.py
-
Run it
- Optional cloud-based deployment using external AI APIs
- Streaming responses (WebSockets)
- UI modernization using React + Tailwind
- Enhanced memory retrieval strategies
Built with ❤️ by Pranziss / yubedaoneineed
This is my first public chatbot project — feel free to fork, star ⭐, or reach out with feedback and ideas.