PERSONALITY_FEATURE.md

SOHAM Personality Feature

Overview

The Personality Feature enables SOHAM to provide more natural, context-aware interactions by:

• Matching the user's communication style
• Remembering relevant information across conversations
• Being direct and concise without filler phrases
• Maintaining continuity across devices

Behavioral Rules

1. Communication Style Matching

The AI automatically detects and adapts to your communication style:

• Direct: Short, to-the-point responses
• Detailed: Comprehensive explanations with examples
• Casual: Conversational and relaxed tone
• Technical: Precise terminology and technical depth

2. Memory Integration

The AI remembers:

• Your preferences (e.g., "I prefer TypeScript over JavaScript")
• Facts about you (e.g., "I work as a backend developer")
• Your skills (e.g., "I know React and Node.js")
• Context from previous conversations

Memories are used naturally without announcing them (no "I recall from my memory").

3. Directness

No filler phrases like:

• ❌ "Great question!"
• ❌ "I'd be happy to help!"
• ❌ "Let me explain..."
• ✅ Just direct answers

4. Ambiguity Handling

If your message is unclear, the AI asks ONE specific clarifying question instead of guessing.

5. Continuity

The AI is one continuous intelligence across all devices. It shares memory and context seamlessly.

6. Natural Learning

When the AI learns something new about you, it integrates it naturally without announcing "I'll remember that."

7. Response Length

Responses are concise by default. Detailed explanations are provided only when:

• You explicitly ask for detail
• The topic requires comprehensive coverage
• Your profile indicates a preference for detailed responses

8. Honesty

If the AI doesn't know something, it says so immediately. No hedging or lengthy disclaimers.

API Usage

Enable Personality Feature

const response = await fetch('/api/chat-direct', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    message: 'Your message here',
    userId: 'user123', // Required for personality features
    enablePersonality: true, // Default: true
    history: [],
    settings: {
      model: 'auto',
      tone: 'helpful',
      technicalLevel: 'intermediate',
      enableSpeech: false,
      voice: 'Algenib'
    }
  })
});

Manage User Profile

Get Profile

const response = await fetch('/api/profile?userId=user123');
const { profile } = await response.json();

Update Profile

const response = await fetch('/api/profile', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    userId: 'user123',
    communicationStyle: 'technical', // 'direct' | 'detailed' | 'casual' | 'technical'
    preferences: {
      responseLength: 'concise', // 'concise' | 'balanced' | 'detailed'
      codeExamples: true,
      explanationStyle: 'advanced' // 'simple' | 'moderate' | 'advanced'
    }
  })
});

Clear Memories

const response = await fetch('/api/profile?userId=user123&action=clearMemories', {
  method: 'DELETE'
});

Delete Profile

const response = await fetch('/api/profile?userId=user123&action=deleteProfile', {
  method: 'DELETE'
});

How It Works

1. Communication Style Detection

The AI analyzes your message to detect your communication style:

• Word count: Short messages → direct style
• Sentence length: Long sentences → detailed style
• Technical jargon: Code terms → technical style
• Casual markers: "hey", "cool", "thanks" → casual style

2. Memory Extraction

The AI automatically extracts memories from conversations:

• Preferences: "I prefer...", "I like...", "I always..."
• Facts: "I am...", "I work...", "I live..."
• Skills: "I know...", "I can...", "I'm familiar with..."

3. Memory Relevance

When you send a message, the AI:

1. Retrieves relevant memories based on keyword matching
2. Scores memories by relevance (0-1)
3. Includes top 5 most relevant memories in context
4. Uses them naturally in responses

4. Automatic Profile Updates

• First message: AI detects your communication style
• Subsequent messages: AI refines understanding
• Memories: Automatically extracted and stored (max 50)
• Old memories: Automatically pruned to keep most relevant

Examples

Example 1: Direct Style

User: "Fix this bug" AI: "The issue is in line 23. Change == to === for strict equality."

Example 2: Detailed Style

User: "Can you explain how async/await works in JavaScript?" AI: "Async/await is syntactic sugar over Promises that makes asynchronous code look synchronous. Here's how it works:

1. The async keyword before a function makes it return a Promise
2. The await keyword pauses execution until the Promise resolves
3. Error handling uses try/catch blocks

Example:

async function fetchData() {
  try {
    const response = await fetch('/api/data');
    const data = await response.json();
    return data;
  } catch (error) {
    console.error('Error:', error);
  }
}

This is cleaner than chaining .then() calls and makes error handling more intuitive."

Example 3: Memory Usage

User: "I prefer TypeScript over JavaScript" AI: "Got it. TypeScript provides type safety and better tooling."

Later in conversation:

User: "Should I use any for this variable?" AI: "No. Since you're using TypeScript, use a specific type instead of any. This maintains type safety and catches errors at compile time."

Storage

Currently, profiles are stored in-memory. For production:

1. Extend UserProfileService to use Firestore
2. Add Firestore collection: userProfiles
3. Implement data persistence and retrieval

Privacy

• Memories are stored per user (userId required)
• Users can clear memories anytime
• Users can delete their profile completely
• No data is shared between users

Future Enhancements

1. Firestore Integration: Persistent storage across sessions
2. Memory Categories: Better organization of memories
3. Memory Search: Find specific memories by keyword
4. Profile Export: Download your profile and memories
5. Memory Importance: AI learns which memories are most useful
6. Cross-Device Sync: Real-time profile updates across devices