Implementation:PrefectHQ Prefect DataAnalysis Output Model

Metadata
Source	Repo: Prefect
Domains	AI_Agents, Data_Validation
Last Updated	2026-02-09 00:00 GMT

Overview

Concrete Pydantic model defining the structured output schema for the AI data analyst agent in Prefect workflows.

Description

The DataAnalysis BaseModel defines the schema for AI agent output with four fields: summary (str), key_findings (list[str] with min/max length), recommendations (list[str] with min/max length), and columns_analyzed (list[str]). It includes a custom __str__ method for formatted display. This is a Pattern Doc since it defines a user-created schema rather than a library API.

Code Reference

• Repository: https://github.com/PrefectHQ/prefect
• File: examples/ai_data_analyst_with_pydantic_ai.py (L123-164)
• Signature:

class DataAnalysis(BaseModel):
    """Structured analysis results from the AI agent."""
    summary: str = Field(description="High-level summary of the dataset")
    key_findings: list[str] = Field(
        description="Key findings discovered from the data",
        min_length=3, max_length=5
    )
    recommendations: list[str] = Field(
        description="Actionable recommendations based on the findings",
        min_length=3, max_length=5,
    )
    columns_analyzed: list[str] = Field(
        description="List of columns that were analyzed"
    )

• Import: from pydantic import BaseModel, Field

I/O Contract

Direction	Name	Type	Description
Input	(LLM output)	JSON	LLM generates JSON matching this schema
Output	summary	str	High-level summary of the dataset
Output	key_findings	list[str] (3-5 items)	Key findings discovered from the data
Output	recommendations	list[str] (3-5 items)	Actionable recommendations based on findings
Output	columns_analyzed	list[str]	List of columns that were analyzed

Usage Example

from pydantic import BaseModel, Field
from pydantic_ai import Agent

class DataAnalysis(BaseModel):
    summary: str = Field(description="High-level summary")
    key_findings: list[str] = Field(min_length=3, max_length=5)
    recommendations: list[str] = Field(min_length=3, max_length=5)
    columns_analyzed: list[str]

agent = Agent("openai:gpt-4o", output_type=DataAnalysis)
result = await agent.run("Analyze the dataset", deps=df)

# result.output is a validated DataAnalysis instance
print(result.output.summary)
print(result.output.key_findings)

Related Pages

• Principle:PrefectHQ_Prefect_Structured_AI_Output