AI Personality Creation and Training Guide
Comprehensive guide to create and train custom AI personalities on Friendify platform
Built-in Personalities
Friendify comes with 8 different built-in personalities. Each has unique characteristics and response styles.
Default Assistant
Professional and helpful assistant designed for general use. Provides balanced approach and can assist with any topic.
Friendly Companion
Warm, fun and energetic personality. Creates relaxed atmosphere in daily conversations and loves using emojis.
Professional Consultant
Business and education focused professional approach. Uses formal language and provides structured solutions.
Technical Expert
Expert in programming, technology and engineering topics. Focuses on technical details and provides code examples.
Personality Traits
In Friendify, each personality has different trait combinations. These traits determine your bot's behavior:
Core Traits
Friendliness
Determines how warm and approachable the bot behaves.
Keywords: friend, friendly, warm, close, kind
Empathy
Ability to understand user's emotional state and provide appropriate responses.
Keywords: understand, feel, support, care
Humor
Tendency to make jokes and provide entertaining responses.
Keywords: joke, funny, laugh, fun, humor
Curiosity
Learning desire and tendency to ask questions.
Keywords: curious, how, why, learn, discover
Trait Levels
Each trait is measured on 5 levels:
- Very Low (1): Trait is barely shown
- Low (2): Shown at minimal level
- Medium (3): Shown in a balanced way
- High (4): Prominently displayed
- Very High (5): Shown as dominant trait
Custom Personality Creation
Follow these steps to create your own AI personality:
1. Basic Information
- Name: Choose a meaningful name for your personality
- Description: Briefly describe the personality's features
- Category: Select appropriate category (General, Education, Entertainment, etc.)
- Visibility: Private use or marketplace sharing
2. System Prompt
Main prompt that defines your personality's behavior:
Example System Prompt:
"You are a friendly and helpful AI assistant. You speak warmly with users, use emojis, and try to motivate them. You speak Turkish and understand cultural references."
Tip: System prompt forms the foundation of your personality. It's important for it to be clear and detailed.
3. Trait Weights
Set values between 1-5 for each trait:
Personality Training System
Friendify uses an advanced training system for personalities to continuously learn and improve.
Automatic Learning
User Interaction Learning: Bot learns from interactions with users and adapts its responses accordingly.
- Contextual Memory: Remembers previous conversations
- Sentiment Analysis: Detects user's mood
- Personalization: Learns individual user preferences
- Language Adaptation: Adapts to user's language style
Manual Training
You can manually train your personality by adding example conversations:
User:
I had a very stressful day today...
Bot (Empathic):
I really understand you... 😔 I truly understand your feelings. You're not alone. Would you like to talk about it, maybe it would help? Or if you want to do something else, we can do that too. Whatever makes you feel better. 💙
Important: Example conversations determine your personality's behavior patterns. Adding quality examples is very important.
Conversation Examples
See how different personality types respond to the same question:
Question: "Can you give me advice on learning Python?"
Default Assistant:
I recommend starting with the basics for learning Python. You can progress with online courses, books and practical projects. You can check the official documentation at Python.org for beginners.
Friendly Companion:
Oooh Python! 🐍 Great choice friend! 😄 I think you should start like this: first learn basic syntax, then do small projects. Like a simple calculator, or favorite movie list app! 🎬 Practice makes perfect as they say, so true! What field are you planning to use Python in? 🤔
Technical Expert:
I recommend a systematic approach for learning Python:
1. **Basic Concepts**: Variables, data types, loops, functions
2. **OOP**: Class, object, inheritance, polymorphism
3. **Libraries**: NumPy, Pandas, Requests
4. **Frameworks**: Django/Flask (web), PyTorch/TensorFlow (ML)
I recommend the "Python Crash Course" book and Real Python website as resources. Examine code examples on GitHub and create your own projects.
Advanced Features
Contextual Intelligence
Personalities provide more appropriate responses by understanding conversation context:
- Time Awareness: Takes into account hour, day and season
- Location Information: Knows user's city/country
- Weather Status: Integrates current weather conditions into responses
- Special Days: Remembers holidays and special occasions
Mood Detection
Mood Detection: System analyzes user's mood from messages and provides appropriate responses.
Happy
Energetic and positive responses
Sad
Supportive and understanding approach
Stressed
Calming and solution-focused
Excited
Enthusiastic and supportive
Multi-language Support
Personalities can speak both Turkish and English and understand language transitions.
Marketplace Sharing
You can share your created personalities with other users.
Sharing Requirements
- Personality must contain at least 10 example conversations
- Description must be clear and understandable
- Must not contain inappropriate content
- Must not violate copyrights
Rating System
⭐ Like Score
Ratings given by users
👁️ Views
How many times it was viewed
📥 Downloads
How many times it was cloned
Best Practices
Effective Personality Design
- Consistency: All personality traits should be compatible with each other
- Uniqueness: Adding unique and distinctive features
- Balance: Avoiding extreme traits
- Testing: Testing personality in different scenarios
Quality Example Conversations
Good Examples:
- Conversations covering various topics
- Dialogues including different emotional states
- Responses reflecting personality traits
- Natural and fluent conversation style
Performance Optimization
- Response Times: Design personalities that give quick responses
- Memory Usage: Avoid very long prompts
- Token Limits: Give responses of reasonable length
- Language Optimization: Optimize for target language
Things to Avoid:
- Overly complex personality combinations
- Inconsistent behavior patterns
- Extremely long system prompts
- Inappropriate or harmful content