Switched the AIs for email generation and revision to llama3.1:8b from Ollama instead of huggingface

docs/ai/OLLAMA_SETUP.md

@@ -0,0 +1,345 @@
+# Ollama Setup for AI Email Generation
+
+This guide will help you set up Ollama with the llama3.1:8b model for AI-powered email generation and revision features in your DripEmails application.
+
+## Overview
+
+The application uses **Ollama** with the **llama3.1:8b** model (8 billion parameter model), which provides:
+
+- High-quality professional email content generation
+- Email revision for grammar, clarity, and tone improvements
+- Personalized email campaign creation
+- **Complete privacy** - runs entirely on your local machine
+- **No API keys required** - no external service dependencies
+- **CPU-only operation** - no GPU required
+
+## Prerequisites
+
+- 8GB RAM minimum (16GB recommended)
+- 8GB free disk space for the model
+- 4+ CPU cores recommended
+- Internet connection for initial model download
+- Windows, macOS, or Linux operating system
+
+## Step 1: Install Ollama
+
+### Windows
+
+1. **Download Ollama**
+
+   - Visit [https://ollama.ai](https://ollama.ai)
+   - Click "Download for Windows"
+   - Run the installer (OllamaSetup.exe)
+
+2. **Verify Installation**
+   - Open PowerShell
+   - Run: `ollama --version`
+   - You should see the Ollama version number
+
+### macOS
+
+1. **Download Ollama**
+
+   - Visit [https://ollama.ai](https://ollama.ai)
+   - Click "Download for Mac"
+   - Open the downloaded .dmg file
+   - Drag Ollama to Applications
+
+2. **Verify Installation**
+   - Open Terminal
+   - Run: `ollama --version`
+   - You should see the Ollama version number
+
+### Linux
+
+1. **Install via Script**
+
+   ```bash
+   curl -fsSL https://ollama.ai/install.sh | sh
+   ```
+
+2. **Verify Installation**
+   ```bash
+   ollama --version
+   ```
+
+## Step 2: Download the llama3.1:8b Model
+
+1. **Pull the Model**
+
+   Open your terminal/PowerShell and run:
+
+   ```bash
+   ollama pull llama3.1:8b
+   ```
+
+2. **Wait for Download**
+
+   - The model is approximately 4.7GB
+   - Download time depends on your internet speed
+   - You only need to do this once
+
+3. **Verify Model Installation**
+   ```bash
+   ollama list
+   ```
+   You should see `llama3.1:8b` in the list
+
+## Step 3: Start Ollama with CPU-Only Mode
+
+**IMPORTANT:** To ensure Ollama runs on CPU only (avoiding GPU driver requirements), you must set the `OLLAMA_NUM_GPU=0` environment variable.
+
+### Windows PowerShell
+
+**Temporary (current session only):**
+
+```powershell
+$env:OLLAMA_NUM_GPU=0
+ollama serve
+```
+
+**Create a startup script** (recommended):
+
+1. Create a file named `start-ollama-cpu.bat`:
+
+   ```batch
+   @echo off
+   set OLLAMA_NUM_GPU=0
+   ollama serve
+   pause
+   ```
+
+2. Run this batch file to start Ollama
+
+### macOS/Linux
+
+**Temporary (current session only):**
+
+```bash
+export OLLAMA_NUM_GPU=0
+ollama serve
+```
+
+**Permanent (add to shell profile):**
+
+Add to `~/.bashrc`, `~/.zshrc`, or equivalent:
+
+```bash
+export OLLAMA_NUM_GPU=0
+```
+
+Then run:
+
+```bash
+source ~/.bashrc  # or ~/.zshrc
+ollama serve
+```
+
+**Create a startup script** (recommended):
+
+1. Create `start-ollama-cpu.sh`:
+
+   ```bash
+   #!/bin/bash
+   export OLLAMA_NUM_GPU=0
+   ollama serve
+   ```
+
+2. Make it executable:
+
+   ```bash
+   chmod +x start-ollama-cpu.sh
+   ```
+
+3. Run it:
+   ```bash
+   ./start-ollama-cpu.sh
+   ```
+
+## Step 4: Test the Setup
+
+1. **Keep Ollama Running**
+
+   - Leave the `ollama serve` command running in a terminal/PowerShell window
+   - You should see output like: "Ollama is running"
+
+2. **Test in Another Terminal/PowerShell Window**
+
+   ```bash
+   curl http://localhost:11434/api/generate -d '{
+     "model": "llama3.1:8b",
+     "prompt": "Write a professional welcome email",
+     "stream": false
+   }'
+   ```
+
+3. **Expected Response**
+   - You should receive a JSON response with generated text
+   - Response time: 10-30 seconds on CPU (this is normal)
+   - If you get a connection error, ensure `ollama serve` is running
+
+## Step 5: Configure DripEmails (Optional)
+
+If you're running Ollama on a different host or port, create a `.env` file in your project root:
+
+```env
+# Ollama Configuration (defaults shown)
+OLLAMA_BASE_URL=http://localhost:11434
+OLLAMA_MODEL=llama3.1:8b
+```
+
+**Note:** These are the default values. You only need to set them if you're using different settings.
+
+## Performance Expectations
+
+### CPU-Only Mode
+
+- **Generation Time:** 10-30 seconds per email (normal for CPU)
+- **Quality:** Excellent - llama3.1:8b provides high-quality output
+- **Memory Usage:** ~5GB RAM during inference
+- **CPU Usage:** High during generation, minimal when idle
+
+### Tips for Best Performance
+
+1. **Close unnecessary applications** to free up CPU and RAM
+2. **Be patient** - CPU inference takes time but produces quality results
+3. **Keep Ollama running** in the background to avoid startup delays
+4. **Use the startup scripts** to ensure correct CPU-only configuration
+
+## Troubleshooting
+
+### Ollama Won't Start
+
+**Issue:** `ollama serve` fails or exits immediately
+
+**Solutions:**
+
+- Check if another instance is already running: `ps aux | grep ollama` (Linux/Mac) or Task Manager (Windows)
+- Kill existing instances and try again
+- Check port 11434 is not in use: `netstat -an | grep 11434`
+
+### Model Not Found
+
+**Issue:** Error: "model 'llama3.1:8b' not found"
+
+**Solutions:**
+
+- Run: `ollama pull llama3.1:8b`
+- Verify with: `ollama list`
+- Ensure you have enough disk space (8GB required)
+
+### Connection Refused
+
+**Issue:** "Connection refused" error when DripEmails tries to connect
+
+**Solutions:**
+
+- Ensure `ollama serve` is running
+- Check the correct port: http://localhost:11434
+- Verify firewall isn't blocking the connection
+- Test with curl as shown in Step 4
+
+### Slow Generation (>60 seconds)
+
+**Issue:** AI generation takes more than 60 seconds
+
+**Solutions:**
+
+- **Expected on slower CPUs** - 10-30 seconds is normal, up to 60 is acceptable
+- Close other CPU-intensive applications
+- Consider upgrading RAM to 16GB+
+- Ensure OLLAMA_NUM_GPU=0 is set (GPU initialization can slow things down)
+
+### Out of Memory Errors
+
+**Issue:** Ollama crashes or returns memory errors
+
+**Solutions:**
+
+- Ensure you have at least 8GB RAM
+- Close other applications to free memory
+- Restart Ollama service
+- Consider upgrading to 16GB RAM for better stability
+
+### Wrong GPU Being Used
+
+**Issue:** Ollama tries to use GPU despite CPU-only setting
+
+**Solutions:**
+
+- **Critical:** Ensure `OLLAMA_NUM_GPU=0` is set BEFORE starting ollama
+- Restart the terminal/PowerShell and set the variable again
+- Use the startup scripts provided in Step 3
+- Verify with environment variable check:
+  - Windows: `echo $env:OLLAMA_NUM_GPU`
+  - Linux/Mac: `echo $OLLAMA_NUM_GPU`
+
+## Security Notes
+
+- **Privacy:** All AI processing happens locally on your machine
+- **No API Keys:** No external service accounts required
+- **Network:** Ollama only needs internet for initial model download
+- **Data:** Your email content never leaves your machine
+
+## Updating Ollama
+
+To update Ollama to the latest version:
+
+### Windows
+
+- Download the latest installer from https://ollama.ai
+- Run the installer (it will update automatically)
+
+### macOS
+
+- Download the latest version from https://ollama.ai
+- Replace the old application
+
+### Linux
+
+```bash
+curl -fsSL https://ollama.ai/install.sh | sh
+```
+
+## Alternative Models
+
+While DripEmails is configured for `llama3.1:8b`, you can experiment with other models:
+
+### Smaller (faster, less RAM)
+
+```bash
+ollama pull llama3.1:7b   # Similar to 8b but slightly smaller
+```
+
+### Larger (better quality, more resources)
+
+```bash
+ollama pull llama3.1:13b  # Requires 16GB+ RAM
+```
+
+To use a different model, update your `.env` file:
+
+```env
+OLLAMA_MODEL=llama3.1:13b
+```
+
+**Note:** CPU generation time increases with larger models.
+
+## Getting Help
+
+- **Ollama Documentation:** https://github.com/ollama/ollama/blob/main/README.md
+- **DripEmails Issues:** See the project's GitHub repository
+- **Model Information:** https://ollama.ai/library/llama3.1
+
+## Summary Checklist
+
+- [ ] Ollama installed and `ollama --version` works
+- [ ] Model downloaded with `ollama pull llama3.1:8b`
+- [ ] OLLAMA_NUM_GPU=0 environment variable set
+- [ ] `ollama serve` running successfully
+- [ ] Tested with curl command (Step 4)
+- [ ] DripEmails can generate emails (10-30 second response time)
+
+---
+
+**Congratulations!** Your AI email generation features are now fully configured and running locally on your machine with CPU-only operation.

test_ai_generation.py

@@ -0,0 +1,35 @@
+"""
+Test script for AI email generation
+"""
+import os
+import sys
+import django
+
+# Setup Django
+sys.path.insert(0, os.path.dirname(__file__))
+os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'dripemails.settings')
+django.setup()
+
+from campaigns.ai_utils import generate_email_content
+
+# Test generation
+print("Testing AI email generation...")
+print("=" * 60)
+
+result = generate_email_content(
+    subject="Welcome to our new product",
+    recipient="new customers",
+    tone="friendly",
+    length="medium",
+    context="We're launching a new fitness tracker called FitBand Pro"
+)
+
+print("\n✓ Generation Result:")
+print(f"Success: {result.get('success')}")
+print(f"\nSubject: {result.get('subject')}")
+print(f"\nBody HTML (first 500 chars):")
+print(result.get('body_html', '')[:500])
+print("\n" + "=" * 60)
+
+if result.get('error'):
+    print(f"\n✗ Error: {result.get('error')}")