Agno Agents, Teams and Workflows support various forms of input and output, from simple string-based interactions to structured data validation using Pydantic models.
Structured Output
Generate validated, type-safe outputs using Pydantic models for reliable production systems. Structured outputs provide consistent, predictable response formats instead of unstructured text.
Common use cases include data extraction, classification, content generation, and validation.
Structured outputs work seamlessly with tools, knowledge bases, and most other Agno features.
Agent Example
from pydantic import BaseModel, Field
from agno.agent import Agent
from agno.models.openai import OpenAIChat
class MovieScript(BaseModel):
setting: str = Field(..., description="Setting for the movie")
genre: str = Field(..., description="Genre of the movie")
storyline: str = Field(..., description="Storyline for the movie")
agent = Agent(
model=OpenAIChat(id="gpt-5-mini"),
description="You write movie scripts.",
output_schema=MovieScript,
)
response = agent.run("Write a movie script about New York")
print(response.content) # Returns MovieScript object
Team Example
from pydantic import BaseModel
from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.team import Team
class StockReport(BaseModel):
symbol: str
company_name: str
analysis: str
team = Team(
model=OpenAIChat(id="gpt-5-mini"),
members=[...],
output_schema=StockReport,
)
response = team.run("Analyze NVDA stock")
print(response.content) # Returns StockReport object
Workflow Example
from pydantic import BaseModel
from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.workflow import Step, Workflow
class ResearchFindings(BaseModel):
summary: str
insights: list[str]
research_agent = Agent(
model=OpenAIChat(id="gpt-5-mini"),
output_schema=ResearchFindings,
)
workflow = Workflow(
steps=[Step(agent=research_agent)],
)
response = workflow.run("Research AI trends")
print(response.content) # Returns ResearchFindings object
For models that don’t support structured output, Agno provides JSON mode, which instructs the model to respond in JSON format. While less accurate than native structured output, JSON mode offers a fallback option for compatible models. Set use_json_mode=True on your Agent, Team, or Workflow.
TypedDict. Pass them to the input parameter in run() or print_response().
Agent Example
from pydantic import BaseModel, Field
from agno.agent import Agent
from agno.models.openai import OpenAIChat
class ResearchTopic(BaseModel):
topic: str
focus_areas: list[str] = Field(description="Areas to focus on")
sources_required: int = Field(default=5)
agent = Agent(
model=OpenAIChat(id="gpt-5-mini"),
tools=[...],
)
agent.run(
input=ResearchTopic(
topic="AI",
focus_areas=["Machine Learning", "LLMs"],
sources_required=5,
)
)
Team Example
from pydantic import BaseModel, Field
from agno.team import Team
from agno.models.openai import OpenAIChat
class ResearchProject(BaseModel):
project_name: str
research_topics: list[str] = Field(min_items=1)
max_sources: int = Field(ge=3, le=20, default=10)
team = Team(
model=OpenAIChat(id="gpt-5-mini"),
members=[...],
)
team.run(
input=ResearchProject(
project_name="AI Framework Comparison",
research_topics=["LangChain", "CrewAI", "Agno"],
max_sources=15,
)
)
Workflow Example
from pydantic import BaseModel, Field
from agno.workflow import Workflow
class ResearchRequest(BaseModel):
topic: str = Field(description="Research topic")
depth: int = Field(description="Research depth (1-10)")
workflow = Workflow(
steps=[...],
input_schema=ResearchRequest,
)
workflow.run(
input=ResearchRequest(
topic="AI trends 2024",
depth=8,
)
)
Typesafe Patterns
When you combine both input_schema and output_schema, you create a typesafe component with end-to-end type safety—a fully validated data pipeline from input to output.
Advanced Features
- Parser Model: Use a different model to parse and structure output from your primary model
- Output Model: Use a different model to produce the final output
- Input Validation: Set
input_schema to automatically validate dictionary inputs
- Streaming: Structured outputs work seamlessly with streaming
Learn More
