Alpha release. Install with pip install adx. APIs may change before v1.0. Roadmap →
Skip to content

Examples

Real-world document processing workflows using ADX.

Invoice Extraction

Extract vendor, line items, totals, and tax from PDF invoices. Validate arithmetic and cite every field.

Contract Review

Pull key terms from contracts: parties, dates, governing law, termination clauses. Navigate by section.

Financial Model

Inspect Excel financial models: trace formulas, read assumptions, verify calculations across sheets.

Quick Patterns

Upload and Profile

python
from adx import ADX

dn = ADX()
doc_id = dn.upload("document.pdf")
profile = dn.profile(doc_id)
print(f"Type: {profile['document_type']}, Tables: {profile['table_count']}")

Extract and Validate

python
extraction = dn.extract(doc_id, schema="invoice")
validation = dn.validate(doc_id, extraction["id"])
errors = [c for c in validation["checks"] if c["severity"] == "error"]
if errors:
    print(f"Found {len(errors)} validation errors")

Search and Drill Down

python
hits = dn.search(doc_id, query="payment terms")
for hit in hits["matches"]:
    page = dn.get_page(doc_id, hit["page"])
    # process page content

Documents are not text blobs. Give your agents document tools.