14 KiB
14 KiB
AI Assistant Tutorial
Learn how to integrate AeThex's AI assistant into your game to provide contextual help and coding assistance to players.
What You'll Build
A game with an in-game AI assistant that can:
- Answer questions about game mechanics
- Provide hints and tips
- Generate code snippets
- Offer context-aware help
Time: 20 minutes
Difficulty: Beginner
Prerequisites: Basic GDScript knowledge
Prerequisites
- Completed First Game Tutorial
- AeThex Cloud connection set up
- Basic understanding of UI systems
Step 1: Connect to AI Service
First, ensure you're connected to AeThex Cloud with AI services enabled.
# main.gd
extends Node
func _ready():
# Connect to cloud
var result = await AeThexCloud.connect_to_cloud()
if result.success:
print("Connected to AeThex Cloud")
else:
print("Failed to connect: ", result.error)
Step 2: Create the AI Assistant UI
Create a simple chat interface for the AI assistant.
# ai_assistant.gd
extends Control
@onready var chat_history = $VBox/ChatHistory
@onready var input_field = $VBox/HBox/InputField
@onready var send_button = $VBox/HBox/SendButton
@onready var loading_indicator = $VBox/LoadingIndicator
func _ready():
send_button.pressed.connect(_on_send_pressed)
input_field.text_submitted.connect(_on_text_submitted)
loading_indicator.visible = false
func _on_send_pressed():
_send_message(input_field.text)
func _on_text_submitted(text: String):
_send_message(text)
func _send_message(message: String):
if message.strip_edges().is_empty():
return
# Add user message to chat
add_message("You", message, Color.CYAN)
input_field.clear()
# Show loading indicator
loading_indicator.visible = true
send_button.disabled = true
# Ask AI assistant
var response = await AeThexAI.ask_assistant(message)
# Hide loading indicator
loading_indicator.visible = false
send_button.disabled = false
# Add AI response to chat
if response.success:
add_message("AI Assistant", response.answer, Color.GREEN)
else:
add_message("AI Assistant", "Sorry, I couldn't process that. " + response.error, Color.RED)
func add_message(sender: String, text: String, color: Color):
var label = Label.new()
label.text = "[%s]: %s" % [sender, text]
label.modulate = color
label.autowrap_mode = TextServer.AUTOWRAP_WORD_SMART
chat_history.add_child(label)
# Auto-scroll to bottom
await get_tree().process_frame
if chat_history.get_v_scroll_bar():
chat_history.scroll_vertical = chat_history.get_v_scroll_bar().max_value
Step 3: Create the Scene
Create a scene structure for the AI assistant:
AIAssistant (Control)
├── Panel (Panel)
│ └── VBox (VBoxContainer)
│ ├── Title (Label) - "AI Assistant"
│ ├── ChatHistory (ScrollContainer)
│ │ └── MessageContainer (VBoxContainer)
│ ├── LoadingIndicator (Label) - "Thinking..."
│ └── HBox (HBoxContainer)
│ ├── InputField (LineEdit)
│ └── SendButton (Button) - "Send"
Layout Tips:
- Set Panel anchor to center
- ChatHistory should expand vertically
- InputField should expand horizontally
Step 4: Context-Aware Help
Make the AI assistant aware of the player's current context.
# context_aware_assistant.gd
extends Control
var player_context = {
"current_level": 1,
"player_health": 100,
"inventory": [],
"last_checkpoint": "start",
}
func ask_with_context(question: String):
# Build context string
var context = "Player context: Level %d, Health: %d, Checkpoint: %s" % [
player_context.current_level,
player_context.player_health,
player_context.last_checkpoint
]
# Include context in the question
var full_question = "%s\n\nContext: %s" % [question, context]
# Ask AI
var response = await AeThexAI.ask_assistant(full_question)
return response
func update_context(key: String, value):
player_context[key] = value
Step 5: Pre-defined Quick Help
Add quick help buttons for common questions.
# quick_help.gd
extends Control
@onready var assistant = get_node("../AIAssistant")
var quick_questions = [
"How do I jump higher?",
"What does this item do?",
"Where should I go next?",
"How do I defeat this enemy?",
]
func _ready():
# Create buttons for quick questions
for question in quick_questions:
var button = Button.new()
button.text = question
button.pressed.connect(_on_quick_question.bind(question))
add_child(button)
func _on_quick_question(question: String):
assistant.send_message(question)
Step 6: Code Generation
Use the AI to generate code snippets for players.
# code_generator.gd
extends Control
@onready var code_output = $VBox/CodeOutput
@onready var request_input = $VBox/RequestInput
@onready var generate_button = $VBox/GenerateButton
func _ready():
generate_button.pressed.connect(_on_generate_pressed)
func _on_generate_pressed():
var request = request_input.text
if request.is_empty():
return
# Request code generation
var prompt = "Generate GDScript code for: " + request
var response = await AeThexAI.generate_code(prompt)
if response.success:
code_output.text = response.code
# Add syntax highlighting
code_output.syntax_highlighter = GDScriptSyntaxHighlighter.new()
else:
code_output.text = "Error: " + response.error
Step 7: Hint System
Create a progressive hint system using the AI.
# hint_system.gd
extends Node
var current_puzzle = "temple_door"
var hint_level = 0
func get_hint():
hint_level += 1
var prompt = "Give hint level %d (out of 3) for puzzle: %s. Be progressively more specific." % [
hint_level,
current_puzzle
]
var response = await AeThexAI.ask_assistant(prompt)
if response.success:
return response.answer
else:
return "No hints available"
func reset_hints():
hint_level = 0
Step 8: Tutorial Generator
Let AI generate custom tutorials for players.
# tutorial_generator.gd
extends Node
func generate_tutorial_for_mechanic(mechanic: String):
var prompt = """
Create a brief in-game tutorial for the mechanic: %s
Format:
1. Title
2. 3 simple steps
3. Tips
""" % mechanic
var response = await AeThexAI.ask_assistant(prompt)
if response.success:
return parse_tutorial(response.answer)
else:
return null
func parse_tutorial(text: String) -> Dictionary:
# Parse AI response into structured tutorial
return {
"title": "Tutorial",
"steps": text.split("\n"),
"completed": false
}
Step 9: Error Explanation
Help players understand errors in custom scripting.
# error_explainer.gd
extends Node
func explain_error(error_message: String):
var prompt = """
Explain this game scripting error in simple terms and suggest a fix:
Error: %s
Target audience: Beginners
""" % error_message
var response = await AeThexAI.ask_assistant(prompt)
if response.success:
return {
"explanation": response.answer,
"success": true
}
else:
return {
"explanation": "Could not explain error",
"success": false
}
Step 10: Rate Limiting and Caching
Implement rate limiting to prevent API abuse.
# ai_manager.gd
extends Node
var request_cache = {}
var last_request_time = 0
var min_request_interval = 2.0 # seconds
func ask_with_cache(question: String):
# Check cache first
if question in request_cache:
return request_cache[question]
# Rate limiting
var time_since_last = Time.get_ticks_msec() / 1000.0 - last_request_time
if time_since_last < min_request_interval:
var wait_time = min_request_interval - time_since_last
await get_tree().create_timer(wait_time).timeout
# Make request
last_request_time = Time.get_ticks_msec() / 1000.0
var response = await AeThexAI.ask_assistant(question)
# Cache successful responses
if response.success:
request_cache[question] = response
return response
func clear_cache():
request_cache.clear()
Complete Example: In-Game Help System
Here's a complete help system implementation:
# help_system.gd
extends Control
@onready var help_panel = $HelpPanel
@onready var chat_container = $HelpPanel/VBox/ChatContainer
@onready var input_field = $HelpPanel/VBox/HBox/Input
@onready var send_btn = $HelpPanel/VBox/HBox/SendBtn
@onready var quick_help = $HelpPanel/VBox/QuickHelp
var is_open = false
func _ready():
help_panel.visible = false
send_btn.pressed.connect(_on_send)
input_field.text_submitted.connect(_on_submit)
# Toggle with F1
set_process_input(true)
# Setup quick help buttons
setup_quick_help()
func _input(event):
if event.is_action_pressed("ui_help"): # F1
toggle_help()
func toggle_help():
is_open = !is_open
help_panel.visible = is_open
if is_open:
input_field.grab_focus()
func _on_send():
_send_message(input_field.text)
func _on_submit(text: String):
_send_message(text)
func _send_message(message: String):
if message.strip_edges().is_empty():
return
add_message("You", message, Color.CYAN)
input_field.clear()
# Get context from game
var context = get_game_context()
var full_message = "%s\n\nGame context: %s" % [message, context]
# Ask AI
var response = await AeThexAI.ask_assistant(full_message)
if response.success:
add_message("Assistant", response.answer, Color.GREEN)
else:
add_message("Assistant", "Error: " + response.error, Color.RED)
func add_message(sender: String, text: String, color: Color):
var msg = RichTextLabel.new()
msg.bbcode_enabled = true
msg.text = "[color=%s][b]%s:[/b] %s[/color]" % [color.to_html(), sender, text]
msg.fit_content = true
chat_container.add_child(msg)
func get_game_context() -> String:
# Gather relevant game state
var player = get_tree().get_first_node_in_group("player")
if player:
return "Level: %d, Health: %d, Position: %s" % [
player.current_level,
player.health,
player.global_position
]
return "No context available"
func setup_quick_help():
var questions = [
"How do I play?",
"What are the controls?",
"Where should I go?",
"How do I use items?",
]
for q in questions:
var btn = Button.new()
btn.text = q
btn.pressed.connect(func(): _send_message(q))
quick_help.add_child(btn)
Best Practices
1. Provide Context
Always include relevant game state when asking the AI:
var context = "Player is at checkpoint 3, has 50% health"
var question_with_context = "%s\nContext: %s" % [user_question, context]
2. Clear Expectations
Tell players what the AI can and cannot do:
var help_text = """
AI Assistant can help with:
- Game mechanics
- Quest objectives
- Strategy tips
Cannot help with:
- Technical support
- Account issues
"""
3. Rate Limiting
Prevent abuse with rate limiting:
const MAX_REQUESTS_PER_MINUTE = 10
4. Caching
Cache common questions to reduce API calls:
var faq_cache = {
"how to jump": "Press Space or A button",
"how to save": "Game auto-saves at checkpoints",
}
5. Fallback Responses
Always have fallback responses:
if not response.success:
return "Check the tutorial at Main Menu → Help"
Advanced Features
Personality Customization
Give your AI assistant personality:
func ask_with_personality(question: String):
var system_prompt = """
You are a helpful wizard companion in a fantasy RPG.
Speak in a wise, mystical tone.
Be encouraging and friendly.
"""
var response = await AeThexAI.ask_assistant(
question,
{"system_prompt": system_prompt}
)
return response
Multi-Language Support
Support multiple languages:
func ask_in_language(question: String, language: String):
var prompt = "Answer in %s: %s" % [language, question]
return await AeThexAI.ask_assistant(prompt)
Voice Integration
Combine with text-to-speech:
func speak_response(text: String):
# Use DisplayServer TTS if available
if DisplayServer.tts_is_speaking():
DisplayServer.tts_stop()
DisplayServer.tts_speak(text, "en", 50, 1.0, 1.0, 0, true)
Testing
Test your AI assistant:
# test_ai_assistant.gd
extends Node
func _ready():
run_tests()
func run_tests():
print("Testing AI Assistant...")
# Test basic question
var r1 = await AeThexAI.ask_assistant("How do I move?")
assert(r1.success, "Basic question failed")
# Test code generation
var r2 = await AeThexAI.generate_code("Create a jump function")
assert(r2.success and "func" in r2.code, "Code generation failed")
print("All tests passed!")
Troubleshooting
AI not responding:
- Check cloud connection:
AeThexCloud.is_connected() - Verify AI service is enabled in project settings
- Check console for error messages
Slow responses:
- Implement request caching
- Use loading indicators
- Consider reducing context size
Irrelevant answers:
- Provide more specific context
- Use structured prompts
- Implement feedback system
Next Steps
- Authentication Tutorial - Add user accounts
- Analytics Tutorial - Track AI usage
- API Reference - Complete AI API docs
Summary
You've learned how to:
✅ Integrate AI assistant into your game
✅ Create context-aware help systems
✅ Generate code and tutorials with AI
✅ Implement caching and rate limiting
✅ Build an in-game help UI
The AI assistant can dramatically improve player experience by providing instant, contextual help!
Questions? Ask the AI assistant in your game! 🤖