lasso-ai 0.1.1

Official Python SDK for the Lasso API — AI-powered product data extraction, enrichment, and web search. Extract structured data from PDFs, URLs, and text. Enrich products with AI-generated descriptions, pricing, and specs.

lasso-ai

Official Python SDK for the Lasso API — AI-powered product data extraction, enrichment, and web search.

Search for products across the web and get structured results. Enrich partial product records with complete data including pricing, specs, and descriptions — all backed by citations. Extract structured data from PDFs, spreadsheets, URLs, and raw text into clean, typed tables. Export as JSON, CSV, XLSX, or images.

Features:

• Search — Find products with natural language queries, get structured data back instantly
• Enrich — Pass partial product info, get complete records with citations and confidence scores
• Extract — Pull structured data from PDFs, CSVs, URLs, and free-form text
• Enhance — AI-generate descriptions, translations, and computed fields for any column
• Export — Download tables as JSON, CSV, XLSX, or image archives
• Schemas — Define and reuse column structures across tables
• Glossary — Manage translation terms and brand-specific vocabulary

Installation

pip install lasso-ai

With pandas support:

pip install lasso-ai[pandas]

Usage

from lasso import LassoClient

lasso = LassoClient(api_key="lasso_...")

Schemas

schema = lasso.schemas.create(
    name="Products",
    columns=[
        {"key": "name", "label": "Product Name", "type": "text", "required": True},
        {"key": "price", "label": "Price", "type": "number"},
        {"key": "description", "label": "Description", "type": "richtext"},
    ],
)

schemas = lasso.schemas.list()
detail = lasso.schemas.get(schema["id"])

Files

file = lasso.files.upload("products.pdf")
files = lasso.files.list()

Tables

table = lasso.tables.create(
    schema_id=schema["id"],
    name="My Products",
    file_ids=[file["id"]],
)

# Poll until extraction completes
result = lasso.tables.wait_for_completion(table["id"])

rows = lasso.tables.rows(result["id"])

Enhancement

job = lasso.enhance.cells(
    table["id"],
    row_ids=["row-1", "row-2"],
    column_key="description",
    prompt="Write a product description based on {{Product Name}}",
)

status = lasso.enhance.status(table["id"])

Export

csv_data = lasso.export.csv(table["id"])
json_data = lasso.export.json(table["id"])

# Download xlsx to disk
lasso.export.xlsx(table["id"], path="products.xlsx")

Pandas integration

df = lasso.tables.results_as_dataframe(table["id"])
print(df.head())

Glossary

lasso.glossary.create(term="LCD", type="do_not_translate")
terms = lasso.glossary.list()

Credits

balance = lasso.credits.balance()
usage = lasso.credits.usage(from_="2026-01-01")

Search

results = lasso.search(
    query="wireless noise-cancelling headphones under $200",
    max_results=5,
)

for item in results["results"]:
    print(item["data"]["name"], item["source_url"])

# With a custom schema
results = lasso.search(
    query="ergonomic office chairs",
    columns=[
        {"key": "name", "label": "Name", "type": "text"},
        {"key": "price", "label": "Price", "type": "number"},
        {"key": "rating", "label": "Rating", "type": "number"},
    ],
)

Enrich

enriched = lasso.enrich(
    items=[
        {"data": {"name": "Sony WH-1000XM5"}},
        {"data": {"name": "Apple AirPods Pro 2"}},
    ],
    columns=[
        {"key": "name", "label": "Name", "type": "text"},
        {"key": "price", "label": "Price", "type": "number"},
        {"key": "description", "label": "Description", "type": "richtext"},
    ],
    web_search=True,
    thinking="medium",
)

for item in enriched["items"]:
    print(item["data"]["name"], item["data"]["price"])
    for field in item["basis"]:
        print(f"  {field['field']}: {field['confidence']}", field["citations"])

Context manager

with LassoClient(api_key="lasso_...") as lasso:
    schemas = lasso.schemas.list()

Error handling

from lasso import LassoClient, LassoError

try:
    lasso.schemas.get("non-existent")
except LassoError as e:
    print(e.status_code)   # 404
    print(e.error_type)    # "not_found"
    print(e.request_id)    # "req_..."

License

MIT