Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.agno.com/llms.txt

Use this file to discover all available pages before exploring further.

Turn any input (text, audio, image, video, or file) into structured, reviewed data. Agno is natively multi-modal and type-safe, so a labeling stack is a few lines of Python. Add an output_schema and the agent returns data in that shape:
cookbook/data_labeling/_01_text_classification/basic.py
from typing import Literal

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from pydantic import BaseModel, Field


class Classification(BaseModel):
    label: Literal["positive", "negative", "neutral"] = Field(
        ..., description="The assigned sentiment label"
    )


agent = Agent(
    model=OpenAIResponses(id="gpt-5.5"),
    instructions="You classify product reviews by sentiment.",
    output_schema=Classification,
)

result = agent.run("Broken on arrival, total waste of money.").content
# Classification(label='negative')
The output is a valid Classification object. Swap the schema and instructions and the same pattern covers extraction, span labeling, scoring, and preference ranking.

​
Data labeling workflows

Pick the page that matches the shape of label you need.
WorkloadInputOutputPage
Structured extractionAny modalityTyped Pydantic objectStructured extraction
ClassificationAny modalityOne label, label set, or spansClassification
Scoring / evaluationPrompt + responseRubric scoresLLM as judge
Preference rankingPrompt + two responsesWinner + rationalePreference data
Non-text inputImage, audio, video, PDFAny of the aboveMultimodal inputs
Reviewed labelsAny inputAdjudicated label + audit trailQuality pipeline

​
Guided paths

You haveYou wantStart with
Free-form text or documentsTyped records for a warehouseStructured extraction
Model outputs to gradeA score per outputLLM as judge
Response pairsRLHF / DPO training dataPreference data
Quality requirementsTwo-model agreement and adjudicationQuality pipeline

​
Model choice

Each workload uses the model that fits the modality. For example:
ModalityDefault in the cookbook
Text, document/PDFopenai:gpt-5.5
Imageopenai:gpt-5.5
Audio, videogemini-3-flash-preview
Second labeler in the quality pipelineclaude-sonnet-4-5

​
Explore

Structured extraction

Turn any modality into a typed object, with optional per-field confidence.

Classification

Single-label, multi-label, hierarchical, and span labeling.

LLM as judge

Score outputs against a rubric. The same machinery, used for evals.

Preference data

Rank A vs B for RLHF and DPO datasets.

Multimodal inputs

Feed images, audio, video, and PDFs into any labeler.

Quality pipeline

Two labelers, a reviewer, and an adjudicator with an audit trail.

​
Developer Resources