Skip to main content
AgentCanvas

AI Integration

Connect to popular AI services like ChatGPT (GPT-4, GPT-4o), Claude, Gemini, and others. Each AI service offers different models with varying capabilities and costs. Add your API keys in settings, then use AI nodes in your workflows.

Supported AI Services

AgentCanvas integrates with major AI providers, each offering different models:

OpenAI (ChatGPT)

  • GPT-4 - Most capable, higher cost
  • GPT-4 Turbo - Fast and efficient
  • GPT-4o - Latest optimized model
  • GPT-3.5 Turbo - Cost-effective for simple tasks
  • GPT-4 Vision - Image understanding capabilities

Anthropic (Claude)

  • Claude 3 Opus - Most powerful
  • Claude 3 Sonnet - Balanced performance
  • Claude 3 Haiku - Fastest and most cost-effective

Google (Gemini)

  • Gemini 1.5 Pro - Latest generation
  • Gemini Pro - Standard model

Each model has different strengths. GPT-4 is great for complex reasoning, Claude excels at long-form content, and Gemini offers strong multimodal capabilities. Choose the model that best fits your needs and budget.

Adding API Keys

Before you can use AI services in your workflows, you need to add API keys:

  1. Go to your account settings
  2. Navigate to the API Keys section
  3. Click "Add API Key" and select the service (OpenAI, Anthropic, Google, etc.)
  4. Paste your API key from the provider's dashboard
  5. Give it a name to identify it later
  6. Save—the key is stored securely and encrypted

API keys are stored securely and encrypted. You can add multiple keys for the same service if needed, or use different keys for different workflows.

To get an API key, visit the provider's website (OpenAI, Anthropic, Google) and sign up for an account. Most providers offer free credits to get started.

Using AI Nodes

Once you've added API keys, you can use AI nodes in your workflows:

  • LLM Chat: Chat with AI models—generate text, answer questions, analyze data
  • Image Generation: Create images with DALL-E or other image AI models
  • Speech to Text: Convert audio to text using Whisper
  • Agentic AI: Autonomous AI agent that can use other nodes as tools

Each AI node has settings you can configure:

  • Model selection: Choose which model to use from the provider
  • API key: Select which API key to use (if you have multiple)
  • System prompt: Instructions for the AI—use templates with {{variables}} to reference workflow data
  • Temperature: Control randomness (0 = deterministic, 2 = very creative)
  • Max tokens: Limit how long the response can be

Costs and Billing

The system automatically calculates costs based on your AI usage:

  • Different models have different costs per token
  • The workflow designer shows estimated costs before you run anything
  • You're charged credits based on the all-inclusive rate (base + platform fee + Adyen fees)
  • Costs vary by model—GPT-4 costs more than GPT-3.5, Claude Opus costs more than Haiku

You can see real-time cost estimates as you build your workflow. The system tracks token usage and calculates credits based on the actual provider costs plus platform fees.

The pricing page shows the all-inclusive cost for each AI operation type. These prices include platform fees and payment processing fees, so what you see is what you pay.

Best Practices

  • Choose the right model: Use GPT-3.5 for simple tasks, GPT-4 for complex reasoning
  • Set temperature appropriately: Lower (0-0.7) for factual tasks, higher (0.7-1.5) for creative work
  • Use templates: Reference data from earlier steps using {{variable}} placeholders
  • Limit token usage: Set max_tokens to avoid unexpected costs
  • Test before running: Use the cost estimator to see what your workflow will cost
  • Monitor usage: Check your credit balance regularly and set up usage alerts