Pervaxis.Genesis.AIAssistance.AWS 2.0.0

Pervaxis.Genesis.AIAssistance.AWS

AWS Bedrock-based AI assistance provider for the Pervaxis Genesis platform. Supports text generation, embeddings, and image generation using Claude, Titan, and Stable Diffusion models.

Features

• Text Generation: Claude 3.5 Sonnet and Titan text models
• Embeddings: Titan Embeddings for vector generation
• Image Generation: Stable Diffusion XL for AI-generated images
• Model Flexibility: Easy switching between Claude and Titan models
• LocalStack Support: Full local development support
• Type-Safe: Strongly-typed configuration with validation
• Async/Await: Modern async API throughout
• Comprehensive Logging: Structured logging with Microsoft.Extensions.Logging
• Resource Management: Proper IDisposable implementation

Installation

dotnet add package Pervaxis.Genesis.AIAssistance.AWS

Configuration

appsettings.json

{
  "AIAssistance": {
    "Region": "us-east-1",
    "TextModelId": "anthropic.claude-3-5-sonnet-20241022-v2:0",
    "EmbeddingModelId": "amazon.titan-embed-text-v1",
    "ImageModelId": "stability.stable-diffusion-xl-v1",
    "Temperature": 0.7,
    "MaxTokens": 1024,
    "MaxRetries": 3,
    "RequestTimeoutSeconds": 60
  }
}

Dependency Injection

using Pervaxis.Genesis.AIAssistance.AWS.Extensions;

// Option 1: From configuration
services.AddGenesisAIAssistance(configuration.GetSection("AIAssistance"));

// Option 2: Inline configuration
services.AddGenesisAIAssistance(options =>
{
    options.Region = "us-east-1";
    options.TextModelId = "anthropic.claude-3-5-sonnet-20241022-v2:0";
    options.Temperature = 0.7;
    options.MaxTokens = 2048;
});

Usage

Text Generation

using Pervaxis.Core.Abstractions.Genesis.Modules;

public class ContentService
{
    private readonly IAIAssistant _aiAssistant;

    public ContentService(IAIAssistant aiAssistant)
    {
        _aiAssistant = aiAssistant;
    }

    public async Task<string> GenerateBlogPostAsync(string topic)
    {
        var prompt = $"Write a detailed blog post about: {topic}";
        var content = await _aiAssistant.GenerateTextAsync(prompt);
        return content;
    }

    public async Task<string> AnswerQuestionAsync(string question)
    {
        var prompt = $"Answer this question concisely: {question}";
        var answer = await _aiAssistant.GenerateTextAsync(prompt);
        return answer;
    }
}

Embeddings Generation

public class SearchService
{
    private readonly IAIAssistant _aiAssistant;

    public SearchService(IAIAssistant aiAssistant)
    {
        _aiAssistant = aiAssistant;
    }

    public async Task<float[]> GenerateQueryEmbeddingAsync(string query)
    {
        var embedding = await _aiAssistant.GenerateEmbeddingAsync(query);
        // Use embedding for similarity search, clustering, etc.
        return embedding;
    }

    public async Task<Dictionary<string, float[]>> IndexDocumentsAsync(List<string> documents)
    {
        var embeddings = new Dictionary<string, float[]>();
        
        foreach (var doc in documents)
        {
            var embedding = await _aiAssistant.GenerateEmbeddingAsync(doc);
            embeddings[doc] = embedding;
        }
        
        return embeddings;
    }
}

Image Generation

public class ImageService
{
    private readonly IAIAssistant _aiAssistant;

    public ImageService(IAIAssistant aiAssistant)
    {
        _aiAssistant = aiAssistant;
    }

    public async Task<byte[]> GenerateMarketingImageAsync(string description)
    {
        var prompt = $"Professional marketing image: {description}";
        var imageData = await _aiAssistant.GenerateImageAsync(prompt);
        
        // Save to file
        await File.WriteAllBytesAsync("marketing-image.png", imageData);
        
        return imageData;
    }

    public async Task<byte[]> CreateIllustrationAsync(string concept)
    {
        var prompt = $"Artistic illustration of: {concept}, vibrant colors, high detail";
        var imageData = await _aiAssistant.GenerateImageAsync(prompt);
        return imageData;
    }
}

Model Selection

Text Generation Models

Claude 3.5 Sonnet (Recommended)

"TextModelId": "anthropic.claude-3-5-sonnet-20241022-v2:0"

• Best for: Complex reasoning, analysis, content generation
• Context: 200K tokens
• Cost: $3 per 1M input tokens, $15 per 1M output tokens

Claude 3 Opus

"TextModelId": "anthropic.claude-3-opus-20240229"

• Best for: Highest quality outputs, complex tasks
• Context: 200K tokens
• Cost: $15 per 1M input tokens, $75 per 1M output tokens

Titan Text Express

"TextModelId": "amazon.titan-text-express-v1"

• Best for: Cost-effective text generation
• Context: 8K tokens
• Cost: $0.80 per 1M input tokens, $1.60 per 1M output tokens

Embedding Models

Titan Embeddings G1 (Default)

"EmbeddingModelId": "amazon.titan-embed-text-v1"

• Dimensions: 1,536
• Max input: 8,192 tokens
• Cost: $0.10 per 1M tokens

Titan Embeddings V2

"EmbeddingModelId": "amazon.titan-embed-text-v2:0"

• Dimensions: 1,024 (default) or configurable
• Max input: 8,192 tokens
• Cost: $0.02 per 1M tokens

Image Generation Models

Stable Diffusion XL (Default)

"ImageModelId": "stability.stable-diffusion-xl-v1"

• Resolution: 1024x1024
• Cost: $0.04 per image

Configuration Options

IAM Permissions

Minimum Required Permissions

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "bedrock:InvokeModel"
      ],
      "Resource": [
        "arn:aws:bedrock:*::foundation-model/anthropic.claude-*",
        "arn:aws:bedrock:*::foundation-model/amazon.titan-*",
        "arn:aws:bedrock:*::foundation-model/stability.stable-diffusion-*"
      ]
    }
  ]
}

Model-Specific Permissions

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": "bedrock:InvokeModel",
      "Resource": [
        "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-5-sonnet-20241022-v2:0",
        "arn:aws:bedrock:us-east-1::foundation-model/amazon.titan-embed-text-v1",
        "arn:aws:bedrock:us-east-1::foundation-model/stability.stable-diffusion-xl-v1"
      ]
    }
  ]
}

LocalStack Support

Configuration

{
  "AIAssistance": {
    "Region": "us-east-1",
    "UseLocalEmulator": true,
    "LocalEmulatorUrl": "http://localhost:4566",
    "TextModelId": "anthropic.claude-3-5-sonnet-20241022-v2:0"
  }
}

Docker Compose

version: '3.8'
services:
  localstack:
    image: localstack/localstack:latest
    ports:
      - "4566:4566"
    environment:
      - SERVICES=bedrock
      - DEBUG=1
    volumes:
      - "./localstack:/var/lib/localstack"

Note: LocalStack's Bedrock support is limited. For full local testing, consider using mock implementations.

Error Handling

All operations throw GenesisException on errors:

try
{
    var text = await _aiAssistant.GenerateTextAsync(prompt);
}
catch (GenesisException ex)
{
    _logger.LogError(ex, "AI generation failed: {Message}", ex.Message);
    // Handle error appropriately
}

Cost Estimation

Text Generation Example

• Claude 3.5 Sonnet
• 1,000 requests/day
• 500 input tokens per request
• 1,000 output tokens per request

Monthly Cost:

• Input: 30,000 requests × 500 tokens = 15M tokens × $3 = $45
• Output: 30,000 requests × 1,000 tokens = 30M tokens × $15 = $450
• Total: $495/month

Embeddings Example

• Titan Embeddings G1
• 10,000 documents indexed
• 500 tokens per document

One-time Cost:

• 10,000 × 500 = 5M tokens × $0.10 = $0.50

Image Generation Example

• Stable Diffusion XL
• 100 images/day

Monthly Cost:

• 3,000 images × $0.04 = $120/month

Best Practices

1. Temperature Settings

• Creative tasks: 0.7-1.0 (blog posts, stories)
• Factual tasks: 0.0-0.3 (summaries, Q&A)
• Balanced tasks: 0.4-0.6 (general content)

2. Token Management

• Set MaxTokens based on use case
• Short responses: 256-512 tokens
• Medium responses: 1024-2048 tokens
• Long responses: 4096+ tokens

3. Prompt Engineering

// Good: Clear, specific prompt
var prompt = "Write a 3-paragraph product description for a smart thermostat. " +
             "Focus on energy savings and ease of use. " +
             "Target audience: homeowners aged 30-50.";

// Poor: Vague prompt
var prompt = "Write about thermostats";

4. Caching Embeddings

// Cache embeddings to avoid redundant API calls
var cacheKey = $"embedding:{Hash(text)}";
if (!_cache.TryGetValue(cacheKey, out float[] embedding))
{
    embedding = await _aiAssistant.GenerateEmbeddingAsync(text);
    _cache.Set(cacheKey, embedding, TimeSpan.FromDays(30));
}

5. Batch Processing

// Process in batches to manage rate limits
var batches = documents.Chunk(10);
foreach (var batch in batches)
{
    var tasks = batch.Select(doc => _aiAssistant.GenerateEmbeddingAsync(doc));
    var embeddings = await Task.WhenAll(tasks);
    await Task.Delay(TimeSpan.FromSeconds(1)); // Rate limiting
}

Troubleshooting

Model Not Found

Error: ResourceNotFoundException Solution: Verify model access in AWS Bedrock console. Some models require manual enablement.

Throttling Errors

Error: ThrottlingException Solution: Implement exponential backoff and increase MaxRetries.

Timeout Errors

Error: TimeoutException Solution: Increase RequestTimeoutSeconds for image generation (can take 30+ seconds).

Invalid Response Format

Error: GenesisException: Invalid response format Solution: Check model ID matches the expected format (Claude vs Titan).

License

Copyright (C) 2026 Clarivex Technologies Private Limited. All Rights Reserved.

Support

• Documentation: https://clarivex.tech/docs/genesis/aiassistance
• Issues: https://github.com/clarivex/pervaxis-genesis/issues
• Email: support@clarivex.tech