Help & Support

Version: 1.0.1

Build Date: 2025‑11‑02

Author: Patrick Andrew Hartnett

RePhrase for Word is an AI-powered text processing add-in for Microsoft Word. Transform, translate, and enhance your text using multiple AI providers including Azure OpenAI, Claude, Gemini, Grok, DeepSeek and more - all without leaving Word.

1. Configure Your First Connection

1. Click the ⚙️ Settings icon
2. Go to Connections tab
3. Click ➕ Add Connection
4. Select your AI provider (e.g., OpenAI, Claude, Gemini)
5. Enter your API Key
6. Choose a Model
7. Click 💾 Save Connection
8. Click ✓ Set Active

2. Process Your First Text

1. Select text in your Word document
2. Choose a Template from the dropdown (optional) or type your prompt directly in the text box
3. Click Send button (or press Ctrl+Enter)
4. View the result in the conversation below
5. Click 📋 Copy, 🔄 Replace, or ➕ Insert buttons on the result card

3. Using the Chat Interface

RePhrase uses a conversational interface similar to ChatGPT:

• Each conversation shows your selected text and the AI's response together
• Multiple conversations are tracked in the same chat session
• Chat actions - Copy, Replace, Insert, Export are available on each result
• Chat history - Access via the ☰ menu to view, search, or create new conversations
• Delete messages - Click 🗑️ on any result to remove it and its associated request

4. Managing Chats

1. Click the ☰ menu to open the navigation panel
2. View all your chat conversations
3. Click 💬 New conversation to start fresh
4. Search, rename, star, export, or delete conversations
5. Switch between conversations by clicking them

• ✅ Chat Interface - Conversational workflow with message history
• ✅ Multi-Chat Management - Create, organize, and search multiple chat sessions
• ✅ Text Processing - Transform selected text with custom prompts
• ✅ Translation - Multi-language support with terminology management
• ✅ Formatting Preservation - Maintains bold, italic, fonts, colors, tables
• ✅ Smart History Matching - Automatically detects and suggests results from previous processing (exact & fuzzy matching with prompt awareness)
• ✅ Processing History - Track and revisit all text processing operations
• ✅ Usage Statistics - Monitor token usage, model distribution, and activity patterns
• ✅ Comparison Differences - Visual comparison of original vs. processed text (for Translation & Writing templates)
• ✅ Prompt Templates - Reusable templates for common tasks
• ✅ Markdown Rendering - Preview formatted AI responses
• ✅ Export Capabilities - Export chats, history, and settings as JSON
• ✅ Connection Testing - Validate credentials before use

Chat & Reasoning Models

• OpenAI - GPT-4o, GPT-4o-mini, o3-mini (reasoning)
• Azure OpenAI - Enterprise deployments with custom models
• Claude - Claude 3.5 Sonnet, Haiku (via Anthropic)
• Google Gemini - Gemini 2.5 Pro, Flash with thinking mode
• Grok - xAI's latest models
• DeepSeek - Cost-effective alternative
• Cohere - Enterprise AI platform

Translation Services

• DeepL - High-quality neural machine translation

Custom APIs

Connect any REST API with:

• OAuth2 authentication
• Bearer token authentication
• API key authentication
• Custom request templates
• JSONPath response extraction

RePhrase automatically tracks all your text processing operations, allowing you to review, reuse, and learn from previous work.

Accessing History

1. Click the 📜 History icon in the top-right corner
2. Browse your processing history (newest first by default)
3. Click any entry to view full details

What's Tracked

Each history entry includes:

• ✅ Selected Text - Original text with formatting (both raw and rendered)
• ✅ Result - AI-generated output (both raw and rendered)
• ✅ Prompts - System instructions and user prompt used
• ✅ Model Info - Which AI connection and model was used
• ✅ Timestamp - When the processing occurred
• ✅ Terminology - Extracted terms (if applicable)

History Features

• Search & Filter - Find entries by text, prompt, or tag
• Sort Options - Newest first, oldest first, or starred first
• Star Favorites - Mark important entries for quick access
• Tags - Organize entries with custom tags
• Notes - Add personal notes to any entry
• Copy Result - Copy processed text to clipboard
• Insert to Word - Directly insert result into your document
• Reuse Prompts - Load prompts from past entries
• Export - Export individual entries or entire history as JSON

Detail View

Click any history entry to see full details with three tabs:

• Selected Text - View original text in formatted or raw (with tags) view
• Result - View AI output in formatted or raw view
• Prompt - Review system instructions and user prompt

Storage Settings

Configure history limits in Settings > General:

• Choose history size: 50, 100, 200, or 500 entries
• Older entries are automatically removed when limit is reached
• Export history before clearing to preserve important entries

Data Management

• Export All History: Settings > General > Export History
• Import History: Settings > General > Import History
• Clear All History: Settings > General > Clear All History
• Delete Single Entry: View entry details > Delete Entry button

RePhrase uses a chat-based interface where each conversation contains a series of text processing requests and AI responses, similar to ChatGPT or Microsoft Copilot.

Creating Conversations

1. A new chat is automatically created when you first open RePhrase
2. Click ☰ menu > 💬 New conversation to start a fresh chat
3. Each chat maintains its own conversation history

Chat Structure

Each conversation shows:

• Context - Which document you were working in
• Your Message - Shows selected text and prompt used
• AI Response - The processed result with actions (Copy, Replace, Insert)
• See more - Expandable section with comparison diff and terminology (if available)

Navigation Panel

Click the ☰ menu to access:

• Search - Find chats by name or content
• Chat List - All your conversations sorted by recent activity
• Star Favorites - Mark important chats with ⭐
• Rename - Give conversations meaningful names (click ✏️)
• Export - Download individual chats or all chats as JSON
• Delete - Remove conversations you no longer need

Chat Actions

• Copy Result - Copy the AI response with formatting
• Replace - Replace original selection with result
• Insert - Insert result after selection
• Export - Download the full conversation data
• Delete Message - Remove a specific request/response pair

Chat vs History

What's the difference?

• Chats - Conversational view, grouped by session, great for ongoing work
• History - Chronological list of all processing, searchable, with detailed analysis
• Both are tracked - Every chat message is also saved to history
• Use chats for - Active work, reviewing recent conversations
• Use history for - Finding older work, analyzing patterns, detailed search

RePhrase automatically detects when you've selected text that you've processed before, helping you save time by reusing previous results instead of re-processing.

How It Works

1. Select text in your Word document
2. If matches are found in history, a 💡 tab appears on the right side of the screen
3. Click the tab to see matching results from your processing history
4. Click any match to instantly load that result (no re-processing needed!)

Match Types

• ✓ Exact Match - The selected text is identical to a previous entry
  • Shown in green badge
  • 100% text match
• ≈ Fuzzy Match - The selected text is very similar (85%+ match)
  • Shown in gold badge
  • Shows similarity percentage (e.g., "92% match")
  • Useful for slightly modified text or minor edits

Prompt Matching

Each match also shows whether it used the same prompt as your current selection:

• ✓ Same Prompt - White checkmark icon
  • Result was generated with the exact same prompt you're using now
  • Highest priority in the list
  • Most likely to be the result you want
• ⚠️ Different Prompt - Yellow warning icon
  • Result was generated with a different prompt
  • May still be useful, but result might vary if re-processed

Match Priority

Matches are automatically sorted by relevance:

1. Exact + Same Prompt - Perfect match (highest priority)
2. Exact + Different Prompt - Same text, different processing
3. Fuzzy + Same Prompt - Similar text, same instructions
4. Fuzzy + Different Prompt - Similar text, different processing

Match Information

Each match card shows:

• Match Type - Exact or Fuzzy with percentage
• Prompt Match - Whether same prompt was used
• Time Ago - When this was processed (e.g., "2h ago")
• Template Name - If a template was used (e.g., "[TRANSLATE FORMAL]")
• Prompt Preview - First few words of the prompt used
• Result Preview - First few words of the result

Using Matches

• Click "Load Result →" to load the result into the main view
• The result appears immediately (no API call needed)
• You can then Replace or Insert as usual
• The panel closes automatically after loading

Managing the Panel

• Opening: Click the 💡 tab on the right side (shows match count)
• Closing: Click the ✕ button or click outside the panel
• Auto-close: Panel closes when you start processing new text
• Persistence: Tab remains visible until you clear your selection

Real-Time Updates

The match panel updates automatically when you:

• Change your selection in Word
• Switch to a different prompt template
• Type or edit your prompt (after 800ms pause)

Track your text processing activity to understand usage patterns, optimize costs, and identify your most effective workflows.

Accessing Statistics

Click the 📊 Statistics icon in the top-right corner to open the usage dashboard.

What You'll See

• Overview Cards - Total queries, tokens used, avg input length, top model
• Activity Timeline - Daily query volume over time
• Models Used - Distribution of which AI models you use most
• Top Prompts - Your most frequently used prompt templates
• Text Transformations - Average similarity, expansion, and formatting preservation
• Token Economics - Detailed breakdown by model with caching and reasoning metrics

Time Periods

View statistics for: Today, Last 7 Days (default), Last 30 Days, Last 90 Days, or All Time.

Token Tracking

• Prompt Tokens - Input sent to AI
• Completion Tokens - Output received from AI
• 💾 Cached Tokens - Tokens served from cache (when supported)
• 🧠 Reasoning Tokens - Internal thinking tokens (o3-mini, Gemini thinking mode)
• ~ Estimated - DeepL shows estimated tokens based on character count

Export Statistics

Click 💾 Export Statistics to save your usage data as JSON for analysis, record keeping, or backup.

What We Track

Absolutely nothing!

• ✅ No analytics or telemetry
• ✅ No user data collection
• ✅ No server-side logging
• ✅ No third-party tracking

Where Your Data is Stored

All settings are stored locally in your browser:

• Location: Browser's localStorage (IndexedDB)
• What's stored: API keys, connection settings, prompt templates, history
• Accessibility: Only accessible from your browser session
• Persistence: Remains until you clear browser data or uninstall

API Communication

• ✅ All API calls use HTTPS (encrypted connections)
• ✅ Direct communication with AI providers (OpenAI, Gemini, etc.)
• ⚠️ CORS Proxy for some providers (Claude, DeepSeek, DeepL)
  • Hosted on Azure Functions (West Europe)
  • No logging or data retention
  • Passes requests through without modification

Clear Your Settings

If you need to remove all stored data:

1. Go to Settings > General
2. Scroll to bottom and click 🔄 Reset to Defaults
3. Confirm when prompted

⚠️ This will clear all settings and restore default connections and templates.

Table Cell Selection Limitations

Current Behavior:

• Selecting cells across multiple rows: ⚠️ Only last selected row is captured

Workaround:

• Select entire table
• Process one cell at a time
• Copy table contents to regular paragraphs, process, then paste back
• Use Word's built-in table tools to extract text before processing

Technical Note: This is a limitation of the Microsoft Word JavaScript API's getSelection() method when dealing with table cell ranges. We are monitoring Office.js updates for potential solutions.

Complex Formatting Across Large Selections

• Issue: Very large selections (multiple pages) with complex formatting may take longer to process
• Impact: Processing times may increase; some edge-case formatting might not be perfectly preserved
• Workaround: Process content in smaller chunks (1-2 pages at a time)

Highlight Colors

• Issue: Custom or theme-based highlight colors may be converted to standard colors
• Impact: Highlights are preserved but color may slightly differ
• Supported Colors: Yellow, Green, Blue, Red, Pink, Turquoise, and standard Office colors

Browser Compatibility

• Best Performance: Microsoft Edge (Chromium), Google Chrome
• Limited Support: Older browsers or Internet Explorer-based Office clients
• Recommendation: Use Office 365 with modern browser for optimal experience

Connection Issues

• Invalid API Key: Double-check your API key in Settings > Connections
• Rate Limit: Some providers have usage limits. Check your account.
• Timeout: Try a shorter text selection or simpler prompt
• CORS Error: These are handled automatically via our proxy

Add-in Not Loading

• Check your internet connection
• Refresh Word (close and reopen)
• Clear browser cache (Ctrl+Shift+Delete in Windows)
• Try reinstalling the add-in

Text Not Processing

• Ensure text is selected in Word
• Check that a connection is active (green checkmark)
• Verify API key is valid using "Test Connection"
• Check I/O tab for detailed error messages

We'd love to hear from you!

Share your feedback, suggest improvements, or let us know about any issues you encounter. Your input helps make RePhrase for Word better for everyone.

What to Include

When sharing feedback, please include:

• RePhrase for Word version (shown at top of this page)
• What you were trying to do
• What happened vs. what you expected
• Any error messages (from I/O tab if relevant)
• Suggestions for improvement

Feedback Types Welcome

• 🐛 Issue reports
• 💡 Feature requests
• 🎨 UI/UX suggestions
• 📚 Documentation improvements
• ⭐ General feedback

License Terms

RePhrase for Word is proprietary software licensed under custom terms. By using this add-in, you agree to the Proprietary Software License .

Key Points

• ✅ Licensed for personal or internal business use
• ❌ No redistribution, sublicensing, or commercial resale
• ❌ No reverse engineering or derivative works
• ⚠️ Software provided "as is" without warranties

For the complete license terms, including limitations of liability and termination clauses, please review the full license document.

Privacy Policy

For details on how RePhrase for Word handles data and privacy, please see our Privacy Policy .

Third-Party Services

This add-in can connect to external AI providers such as OpenAI, Claude, Gemini, DeepL, and others. Your use of those services is subject to their respective terms of service and privacy policies. You are solely responsible for:

• Obtaining and managing API keys
• Complying with third-party service terms
• Any costs incurred from using these services

The author is not responsible for the actions, policies, or services provided by third parties.

Disclaimer

RePhrase for Word is provided "as is" without warranty of any kind, express or implied. The author shall not be liable for any damages arising from the use of or inability to use this software. Use is at your own risk.

Questions?

For licensing inquiries, please contact: patrick.hartnett@live.com

Copyright © 2025 Patrick Andrew Hartnett – All rights reserved.