Smart AI Assistant

This is a unified task-routing AI assistant built using Streamlit and powered by the phi model running locally with Ollama. It supports three intelligent task modes:

• Q&A
• Summarization
• Roleplay (Doctor, Lawyer, Customer Support Agent)

All tasks are routed dynamically with appropriate prompting under the hood. The app features a sleek UI and intelligent input management.

---

Features

• Clean, minimal interface
• Local LLM integration using Ollama
• Task-aware prompt selection (Q&A, Summarization, Roleplay)
• Role-based dynamic prompts for Roleplay mode
• Chat history display with "You" and "AI Assistant" names
• Input box clears automatically after sending

---

Requirements

• Python 3.8+
• Ollama (running the phi model locally)
• Required Python libraries:

  pip install streamlit requests

---

Supported Tasks

1. Q&A

Ask concise, factual questions. The model will generate brief answers.

Prompt Structure:

Answer the following question concisely.
Question: {user_input}
Answer:

2. Summarization

Paste an article, paragraph, or notes. The assistant will summarize it in bullet points.

Prompt Structure:

Summarize the following article in bullet points.
Article: {user_input}
Summary:

3. Roleplay

Choose a role and talk to the assistant like you're speaking to:

• Doctor
• Lawyer
• Customer Support Agent

Prompt Structure Example (Doctor):

You are an experienced medical doctor specialized in general medicine.
The user asks: '{user_input}'
Doctor:

---

How to Run

1. Ensure Ollama is installed and running.

2. Pull the Phi model:

   ollama pull phi

3. Start your app:

   streamlit run app.py

4. Open in browser: http://localhost:8501/

---

Preview

UI Interface with Q&A option selection

Option for Role appears upon Roleplay selection

Demo chat with the AI model

---

Powered by

• Streamlit
• Ollama
• phi Language Model

---

Project Structure

Task-Based-AI-Assistant/
├── task_based_ai_assistant.py      # Main Streamlit app
├── images                          # Folder containing preview images
    ├── img1.png
    ├── img2.png
    ├── img3.png
├── README.md                       # README file briefing about the app
├── image1.png                      # AI Bot Image used in UI

---

📌 Notes

• Make sure phi is running locally via Ollama API (http://localhost:11434).
• Internet is not required once the model is downloaded.
• Easily extendable to include more tasks or roles.