AI Powered Writer Assistant

Generate compelling titles, descriptions, feedback, and visuals for your articles — all with AI.

---

Overview

Writer Assistant is a lightweight Python web application that leverages AI to help writers create better content. Given a raw article draft, it automatically generates:

• A catchy, SEO-friendly title
• A professional article description
• A custom cover image (via Gemini Image Generation)
• Constructive feedback on writing quality and structure

Built with LangChain, Mistral AI, and Google Gemini this tool is perfect for bloggers, technical writers, and developers who want to refine their content with intelligent AI assistance.

Goal: Help writers improve their craft while automating repetitive tasks like brainstorming titles and generating visuals.

---

Features

Feature	Powered By
Article Title Generation	Mistral AI (creative mode)
SEO-Friendly Description	Mistral AI (zero temperature)
Writing Feedback	Structured output from Mistral AI
Cover Image Generation	Google Gemini 2.0 Flash (image generation)
Real-time Web Interface	Built-in HTTP server with live updates
Responsive UI	Pico CSS framework

---

How It Works

1. User inputs an article in the web form.
2. The app sends the text to generate.py, which uses LangChain to orchestrate multiple LLMs.
3. AI generates:
   • Title (creative)
   • Description (accurate)
   • Feedback (structured)
   • Image (via Gemini)
4. All results are saved to the output/ directory.
5. The web server serves the final output with auto-refresh (every 10 seconds).

The app refreshes the image every time you visit the page to avoid caching issues.

---

Project Structure

writer-assistant/
├── generate.py               # Main AI logic: title, description, image, feedback
├── server.py                 # Simple web server with form handling
├── output/                   # Generated assets (title.txt, description.txt, image.png, feedback.txt)
├── input.txt                 # Temporary input file
├── README.md

---

Prerequisites

• Python 3.10+
• pip package manager
• API keys:
  • Mistral AI (for MISTRAL_API_KEY)
  • Google AI Studio (for GOOGLE_API_KEY)

Important: Never commit your API keys to version control.

---

Setup Instructions

1. Install Dependencies

Manually install the dependencies:

pip install langchain mistralai google-generativeai pydantic scikit-image pillow

Or install them from the requirements.txt file:

pip install -r requirements.txt

2. Set Environment Variables

export MISTRAL_API_KEY="your-mistral-api-key-here"
export GOOGLE_API_KEY="your-google-gemini-api-key-here"

Replace with actual keys from your accounts.

3. Run the Server

python3 server.py

Open your browser and go to:
http://localhost:3050

---

Features

• AI-Powered Article Enhancement
  • Generates catchy, SEO-friendly titles
  • Creates compelling meta descriptions
  • Suggests improvements for writing quality
• Dynamic Cover Image Generation
  • Uses Gemini 2.0 Flash Image Generation for visual appeal
  • Automatically updates on refresh
• Constructive Feedback
  • Structured output from LLM to guide improvement
• Simple Web Interface
  • Clean, responsive UI using Pico CSS
  • Real-time feedback loop

---

Why This Project?

This project was built as to learn and experiment with:

• LangChain – For chaining prompts and handling LLM interactions
• Prompt Engineering – Crafting effective system/user messages
• Structured Output – Using Pydantic models for reliable responses
• Image Generation via LLM – Integrating multimodal AI
• Web Server Integration – Connecting backend logic with frontend

---

Acknowledgements

This project was inspired by tutorials on LangChain and AI-powered content generation. It’s a hands-on learning exercise to understand how to combine LLMs, web servers, and structured output in a practical application.