QuoteCheck

Supplier quotes → extraction → normalization → comparison → exceptions → JSON

QuoteCheck is an Apify Actor that compares supplier quotes and turns messy quote data into a structured, machine-readable comparison.

It supports two input modes:

1. PDF quotes - upload text-based supplier quote PDFs.
2. Structured JSON - send already-extracted quote data directly for API and AI-agent workflows.

QuoteCheck detects differences in:

• prices
• quantities
• missing items
• supplier totals
• commercial terms

It can also calculate like-for-like normalized totals when quoted quantities differ, and flags comparisons that require human review.

For AI agents

Use QuoteCheck when a workflow needs to compare supplier quotes, vendor quotes, RFQs, procurement quotes, purchase quotes, or quotation PDFs and return a deterministic, structured comparison.

Typical agent tasks include:

• compare two or more supplier quotes
• reconcile different product descriptions or SKUs
• detect missing items, price differences, and quantity mismatches
• calculate like-for-like normalized totals
• identify comparisons that require human review
• return machine-readable JSON for the next step in an automation

QuoteCheck is designed to be discovered and called as a small business-operation tool through the Apify MCP server, API, or other agent workflows. Agents can inspect the Actor's input and output schemas before running it.

Discovery terms: supplier quote comparison, vendor quote comparison, RFQ comparison, procurement quote comparison, purchase quote reconciliation, PDF quote comparison, supplier price comparison.

MVP

QuoteCheck accepts two input modes:

1. PDF quotes: upload 2–10 text-based supplier quote PDFs.
2. Structured JSON: send normalized quotes directly for API/agent workflows.

For PDF input, the Actor extracts text, detects common quote line-item rows, normalizes quantities and prices, matches equivalent items across suppliers, calculates totals, and returns machine-readable JSON.

PDF flow

PDF Quote A + PDF Quote B → extract → normalize → match → compare → JSON

The current MVP supports text-based PDFs. Scanned/image-only PDFs are detected and reported as requiring OCR rather than silently producing bad data.

Output

Returns:

• matched item groups
• supplier-level quantities and prices
• missing items/suppliers
• quantity differences
• price differences
• calculated supplier totals
• lowest comparable total
• extraction warnings
• review_required flag
• extraction metadata

Structured JSON example

{
  "quotes": [
    {
      "supplier": "Supplier A",
      "items": [
        {"description": "Dell Latitude 7450", "quantity": 10, "unit_price": 1150}
      ]
    },
    {
      "supplier": "Supplier B",
      "items": [
        {"description": "Latitude 7450 Laptop", "quantity": 10, "unit_price": 1095}
      ]
    }
  ]
}

Design principle

QuoteCheck performs one business operation and returns predictable JSON that another automation or AI agent can consume. It does not scrape websites or generate a narrative report.

Limitations

• PDF extraction relies on text embedded in the PDF.
• Scanned PDFs require an OCR layer, which is the next logical upgrade.
• Table layouts vary, so extraction warnings are surfaced and review_required is set when the result needs human checking.