Skip to main content

​
1. Create a schema

A schema defines the columns you want to extract. 
const schema = await client.schemas.create({
  name: "Product Catalog",
  columns: [
    { key: "product_name", label: "Product Name", type: "text", required: true },
    { key: "price", label: "Price", type: "number" },
    { key: "description", label: "Description", type: "text" },
    { key: "image_url", label: "Image", type: "image" },
  ],
});

console.log(schema.id);
You can also generate a schema automatically from sample data:
const schema = await client.schemas.generate({
  sample_data: "Product: iPhone 15 Pro, Price: $999, Storage: 128GB",
  name: "Smartphones",
});

​
2. Upload a file

Upload a PDF, spreadsheet, or image to extract data from. 
const file = new File([buffer], "catalog.pdf", { type: "application/pdf" });
const uploaded = await client.files.upload(file, "catalog.pdf");

console.log(uploaded.id); // "file_abc123"

​
3. Create a table

Start the extraction by creating a table with your schema and uploaded file. 
const table = await client.tables.create({
  schema_id: schema.id,
  name: "Q1 Product Catalog",
  file_ids: [uploaded.id],
});

console.log(table.id);     // "tbl_..."
console.log(table.status); // "processing"

​
4. Wait for results

The SDK includes a polling helper that waits for extraction to complete. 
const completed = await client.tables.waitForCompletion(table.id, {
  intervalMs: 3000,
  timeoutMs: 300000,
});

console.log(completed.status);     // "completed"
console.log(completed.total_rows); // 42

​
5. Retrieve rows

const rows = await client.tables.rows(completed.id, { limit: 100 });

for (const row of rows.data) {
  console.log(row.data.product_name, row.data.price);
}

​
Next

Enhance with AI

Use AI to generate descriptions, translate content, and enrich your data.