PydanticAI Integration Guide

This guide will walk you through integrating PydanticAI with APIpie, enabling you to build type-safe AI applications with structured outputs.

What is PydanticAI?

PydanticAI is a Python agent framework designed to make it easier to build production grade applications with Generative AI. Created by the team behind Pydantic, it brings the same level of type safety and validation to AI applications that Pydantic brings to Python data validation.

Key features include:

• Type-safe interfaces for working with LLMs
• Structured validation of model outputs
• Support for tools and agents
• Dependency injection for modular design
• Built-in support for popular LLM providers

By connecting PydanticAI to APIpie, you unlock access to powerful language models while maintaining type safety and structured outputs.

---

Integration Steps

1. Create an APIpie Account

• Register here: APIpie Registration
• Complete the sign-up process.

2. Add Credit

• Add Credit: APIpie Subscription
• Add credits to your account to enable API access.

3. Generate an API Key

• API Key Management: APIpie API Keys
• Create a new API key for use with PydanticAI.

4. Install PydanticAI

pip install 'pydantic-ai-slim[openai]'

5. Configure PydanticAI for APIpie

You can use PydanticAI with APIpie through its OpenAI-compatible interface:

from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIModel

model = OpenAIModel(
    "openai/gpt-4o-mini",  # or any other APIpie model
    base_url="https://apipie.ai/api/v1",
    api_key="your-APIpie-key-here",
)

agent = Agent(model)
result = await agent.run("Why is the sky blue?")
print(result)

---

Key Features

• Type Safety: Build AI applications that are statically type-checked
• Structured Outputs: Guarantee that LLM responses follow your defined schemas
• Tool Framework: Create tools that LLMs can use, with automatic parameter validation
• Dependency Injection: Easily provide context and services to your agents
• Python-First Design: Use familiar Python patterns and practices with AI

---

Example Workflows

Application Type	What PydanticAI Helps You Build
Type-Safe Assistants	Assistants that return validated, structured data
Function-Calling Agents	Agents that can invoke functions and validate their parameters
Business Logic Integration	AI systems that safely integrate with your existing systems
Multi-Agent Systems	Complex workflows with multiple specialized agents
Data Processing Pipelines	Pipelines that extract, transform, and validate data with LLMs

---

Using PydanticAI with APIpie

Basic Agent

from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIModel

# Configure the model
model = OpenAIModel(
    "openai/gpt-4o-mini",  
    base_url="https://apipie.ai/api/v1",
    api_key="your-APIpie-key-here",
)

# Create a simple agent
agent = Agent(
    model,
    system_prompt="You are a helpful assistant. Be concise and clear."
)

# Run the agent synchronously
result = agent.run_sync("What are three interesting facts about the moon?")
print(result.output)

Using Structured Output

from pydantic import BaseModel, Field
from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIModel

# Define a structured output model
class MoonFacts(BaseModel):
    fact1: str = Field(description="First interesting fact about the moon")
    fact2: str = Field(description="Second interesting fact about the moon")
    fact3: str = Field(description="Third interesting fact about the moon")

# Configure the model
model = OpenAIModel(
    "openai/gpt-4o-mini",
    base_url="https://apipie.ai/api/v1",
    api_key="your-APIpie-key-here",
)

# Create an agent with structured output
facts_agent = Agent(
    model,
    output_type=MoonFacts,
    system_prompt="You are a space expert."
)

# Run the agent
result = facts_agent.run_sync("Tell me about the moon")

# Access the structured data
print(f"Fact 1: {result.output.fact1}")
print(f"Fact 2: {result.output.fact2}")
print(f"Fact 3: {result.output.fact3}")

---

Troubleshooting & FAQ

• Which models are supported?
  Any model available via APIpie's OpenAI-compatible endpoint.

• How do I persist my API key and endpoint?
  Use environment variables or a .env file to store your credentials securely.

• Can I use streaming with PydanticAI?
  Yes, through the agent.stream() method.

For more information, see the PydanticAI documentation or the GitHub repository.

---

Support

If you encounter any issues during the integration process, please reach out on APIpie Discord for assistance.