Text Generation

AI Dev Kit provides multiple methods for generating text content, each optimized for different use cases.

Available Methods

1. Chat Completions (.GENCompletion())

General-purpose chat and text generation:

string response = await "Explain quantum computing"
    .GENCompletion()
    .SetModel(OpenAIModel.GPT4o)
    .ExecuteAsync();

Best for:

• ✅ Conversational AI

• ✅ General text generation

• ✅ Multi-turn conversations

• ✅ Wide provider support

2. Responses API (.GENResponse())

Most capable text generation with advanced features:

string response = await "Write a technical spec"
    .GENResponse()
    .SetModel(OpenAIModel.GPT4o)
    .ExecuteAsync();

Best for:

• ✅ Complex reasoning tasks

• ✅ Long-form content

• ✅ Tool calling support

• ✅ Latest AI capabilities

3. Code Generation (.GENCode())

Specialized for code generation and refactoring:

string code = await "Create a C# quicksort algorithm"
    .GENCode()
    .ExecuteAsync();

Best for:

• ✅ Code generation

• ✅ Code explanation

• ✅ Refactoring suggestions

• ✅ Bug fixes

4. Structured Output (.GENStruct<T>())

Generate JSON mapped to strongly-typed C# classes:

UserProfile profile = await "Create profile for John Doe, age 30"
    .GENStruct<UserProfile>()
    .ExecuteAsync();

Best for:

• ✅ Parsing AI output into C# objects

• ✅ Data extraction

• ✅ Form filling

• ✅ Type-safe responses

Quick Comparison

Method	Use Case	Provider Support	Complexity
GENCompletion()	General chat	Most providers	Simple
GENResponse()	Advanced features	Limited providers	Advanced
GENCode()	Code generation	Most providers	Simple
GENStruct<T>()	JSON output	Most providers	Medium

Basic Examples

Example 1: Simple Question

string answer = await "What is 2+2?"
    .GENCompletion()
    .ExecuteAsync();

Debug.Log(answer); // "2+2 equals 4."

Example 2: Story Generation

string story = await "Write a short sci-fi story about AI"
    .GENResponse()
    .SetModel(OpenAIModel.GPT4o)
    .SetMaxTokens(500)
    .ExecuteAsync();

Debug.Log(story);

Example 3: Code Generation

string code = await "Create a Unity script that rotates an object"
    .GENCode()
    .ExecuteAsync();

Debug.Log(code);

Example 4: Structured Data

[JsonSchema]
public class Character
{
    public string Name { get; set; }
    public int Level { get; set; }
    public string Class { get; set; }
}

Character hero = await "Generate a level 5 warrior named Aragorn"
    .GENStruct<Character>()
    .ExecuteAsync();

Debug.Log($"{hero.Name} (Lv.{hero.Level} {hero.Class})");

Configuration Options

All text generation methods support common configuration:

string response = await "Your prompt"
    .GENCompletion()
    .SetModel(OpenAIModel.GPT4o)           // Model selection
    .SetTemperature(0.7f)                   // Creativity (0.0-2.0)
    .SetMaxTokens(1000)                     // Max response length
    .SetSystemMessage("You are helpful")   // Context
    .SetTopP(0.9f)                          // Nucleus sampling
    .SetFrequencyPenalty(0.5f)              // Reduce repetition
    .SetPresencePenalty(0.5f)               // Encourage diversity
    .ExecuteAsync();

Next Steps

• Chat Completions - Detailed chat API guide

• Responses API - Advanced features

• Code Generation - Code-specific options

• Structured Output - JSON schema guide