# Document processing (/use-cases/document-processing/overview)



Operations teams use document processing agents to move information from invoices, contracts, forms, and scans into databases, ERPs, and review queues. Agno can parse model output from files and images into Pydantic objects. Workflows add approvals, batch execution, retries, and schedules.

Define a Pydantic schema and pass the document through `File`. Check that the run completed and returned an instance of that schema.

<Note>
  `output_schema` requests structured output and Agno attempts to parse it. A failed or unparseable run can leave `content` as text. Check `RunStatus.completed` and the expected Pydantic type before reading fields, indexing, or writing downstream. These examples stop on failure; an application can instead send the original input to a review queue. Schema validation checks the shape and constraints, so verify extracted facts against the source separately.
</Note>

Before running these examples, create and activate a Python environment, then install the provider and set its key:

```bash
uv venv .venv --python 3.12
source .venv/bin/activate
uv pip install "agno[openai]"
export OPENAI_API_KEY="..."
```

Replace every `https://example.com/...` PDF URL with a real PDF URL accessible to the model provider, or use `File(filepath="...")` for a local PDF. Create the referenced local files before running the example. Extraction quality depends on the document and model; the output comments are illustrative.

```python
from agno.run.base import RunStatus

from typing import List, Optional

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


class LineItem(BaseModel):
    description: str
    quantity: Optional[float] = None
    unit_price: Optional[float] = None
    amount: Optional[float] = None


class Invoice(BaseModel):
    invoice_number: Optional[str] = Field(None, description="As printed on the invoice")
    vendor: Optional[str] = None
    invoice_date: Optional[str] = None
    due_date: Optional[str] = None
    subtotal: Optional[float] = None
    tax: Optional[float] = None
    total: Optional[float] = None
    currency: Optional[str] = Field(None, description="ISO 4217, e.g. USD, EUR")
    lines: List[LineItem] = Field(default_factory=list)


agent = Agent(
    model=OpenAIResponses(id="gpt-5.5"),
    instructions=(
        "Extract invoice fields and line items from the attached PDF. "
        "Use exactly what the document shows. If a field is missing, "
        "leave it null. Do not guess."
    ),
    output_schema=Invoice,
)

result_run = agent.run(
    "Extract the invoice.",
    files=[File(url="https://example.com/invoice-1042.pdf")],
)
if result_run.status != RunStatus.completed or not isinstance(result_run.content, Invoice):
    raise RuntimeError("No validated Invoice; send the input to review before continuing")
result = result_run.content
# Invoice(invoice_number='1042', vendor='Acme Corp', invoice_date='2026-04-12',
#         due_date='2026-05-12', subtotal=1200.0, tax=96.0, total=1296.0,
#         currency='USD', lines=[LineItem(...), LineItem(...)])
```

After the checks above, `result` is an `Invoice`. Apply business validation, such as reconciling totals and checking the vendor, before an `INSERT`, ERP call, or queue message.

## Workloads [#workloads]

| Workload                          | Page                                                                          |
| --------------------------------- | ----------------------------------------------------------------------------- |
| Invoices, receipts, statements    | [Invoices and receipts](/use-cases/document-processing/invoices-and-receipts) |
| Contracts, MSAs, policies         | [Contracts](/use-cases/document-processing/contracts)                         |
| Resumes, applications, KYC intake | [Forms and intake](/use-cases/document-processing/forms-and-intake)           |

## Production concerns [#production-concerns]

| You need to                                    | Page                                                                            |
| ---------------------------------------------- | ------------------------------------------------------------------------------- |
| Process a folder or a queue of documents       | [Batch and durability](/use-cases/document-processing/batch-and-durability)     |
| Schedule a nightly run that retries on failure | [Batch and durability](/use-cases/document-processing/batch-and-durability)     |
| Route low-confidence fields to a human         | [Human routing and eval](/use-cases/document-processing/human-routing-and-eval) |
| Track accuracy against a labeled golden set    | [Human routing and eval](/use-cases/document-processing/human-routing-and-eval) |

## Explore [#explore]

<CardGroup cols="2">
  <Card title="Invoices and receipts" icon="file-invoice-dollar" href="/use-cases/document-processing/invoices-and-receipts">
    Header fields, line items, and the path from PDF to a database row.
  </Card>

  <Card title="Contracts" icon="file-signature" href="/use-cases/document-processing/contracts">
    Parties, dates, and a clause-level breakdown for review queues.
  </Card>

  <Card title="Forms and intake" icon="clipboard-list" href="/use-cases/document-processing/forms-and-intake">
    Extract nested employment, education, skills, and identity fields.
  </Card>

  <Card title="Batch and durability" icon="layer-group" href="/use-cases/document-processing/batch-and-durability">
    Workflows over a folder, background runs, scheduled jobs with retries.
  </Card>

  <Card title="Human routing and eval" icon="user-check" href="/use-cases/document-processing/human-routing-and-eval">
    Confidence-gated approval and accuracy tracking against a golden set.
  </Card>
</CardGroup>

## Developer Resources [#developer-resources]

* [Document extraction cookbook](https://github.com/agno-agi/agno/tree/main/cookbook/data_labeling/_16_document_extraction)
* [Structured output](/input-output/structured-output/agent)
* [Workflows](/workflows/overview)
* [Scheduler cookbook](https://github.com/agno-agi/agno/tree/main/cookbook/05_agent_os/12_scheduler)