okdata.app

Creating Templates

Learn how to create and configure templates for your AI workflows.

# Step 1: Name Your Template

Start by giving your template a descriptive name that clearly identifies its purpose.

Template name input field
Best Practices for Naming
  • Use clear, descriptive names
  • Keep it concise but specific
  • Indicate the template's purpose
Examples
  • "Resume Parser" for job applications
  • "Product Analyzer" for e-commerce
  • "Geo Guesser" for location identification

# Step 2: Add a Description

Provide a clear description that explains what your template does.

Template description input

In our example, "Try to guess the city from a picture" clearly explains the purpose of the Geo Guesser template.

# Step 3: Define the Output Schema

The most critical part of creating a template is defining the expected output structure. This determines what fields will be included in your results.

Output schema configuration

Available Data Types:

  • String - For text values
  • Number - For numeric values
  • Boolean - For true/false values
  • Array - For lists of items
  • Object - For nested data structures

# Step 4: Add Custom Prompts (Optional)

You can enhance your template's performance by adding custom prompts that guide the AI on how to process inputs.

Custom prompt configuration

# Step 5: Select AI Model

Choose which AI model should process your template.

LLM model selection

Common options include:

  • GPT-4o: Best for complex reasoning, multimodal inputs, and high accuracy
  • GPT-3.5 Turbo: Faster and more cost-effective for simpler tasks
  • Claude models: Excellent for document understanding and nuanced tasks

# Step 6: Save and Test

Once you've configured your template, you'll want to test it with sample inputs to validate that it correctly processes them and produces the expected output structure.

Template testing playground

Next, learn how to test and fine-tune your templates to ensure optimal performance. Continue to Testing & Fine-tuning.