AI Prompts for Document Extraction

Extract all structured commercial data from this document strictly conforming to standard accounting schemas. Fields required: - vendor_name: (Full legal business entity name) - invoice_number: (Alphanumeric invoice/receipt identifier) - issue_date: (Format: YYYY-MM-DD) - due_date: (Format: YYYY-MM-DD) - subtotal: (Numeric currency amount without symbol) - tax_rate_percentage: (e.g. 5.0) - total_amount_due: (Final payable amount) - line_items: (Array containing description, unit_price, quantity, and line_total) If any field is missing or unreadable, mark it explicitly as null. Do not hallucinate values.

Act as an expert financial forensic auditor. Analyze the uploaded unstructured bank statement/ledger text. Task: 1. Reconstruct all transaction tables into standard rows with columns: [Date | Description | Transaction Type (Debit/Credit) | Amount | Running Balance]. 2. Standardize all transaction descriptions by stripping merchant terminal codes (e.g., turn "POS 4829 SQ *COFFEE SHOP NYC" into "Coffee Shop"). 3. Flag any duplicate entries or statistical outliers. Output format: Provide a markdown table followed by direct CSV copyable code.

Analyze all uploaded research PDFs in this notebook. Synthesize a comprehensive literature review addressing [Research Question]. Structure your synthesis as follows: 1. Core Consensus: What do all sources agree upon? 2. Methodological Divergences: How do authors differ in their datasets or approaches? 3. Identified Gaps: What questions remain unanswered across all papers? 4. Direct Evidence Matrix: Cite specific source titles and page numbers for every assertion made.

Extract all structured commercial data from this Purchase Order document. Strictly output JSON with the schema: { "po_number": string, "order_date": "YYYY-MM-DD", "buyer": { "company_name": string, "billing_address": string, "shipping_address": string, "tax_id": string }, "vendor": { "name": string, "vendor_code": string }, "payment_terms": string, "line_items": [ { "line_no": number, "sku": string, "part_number": string, "description": string, "quantity": number, "unit_of_measure": string, "unit_price": number, "total_price": number } ], "subtotal": number, "tax": number, "freight_shipping_charge": number, "total_po_value": number } If any SKU is ambiguous, prioritize the manufacturer part number (MPN). Do not round unit prices.

Parse this Bill of Lading (BOL) / Freight Manifest into structured tabular format. Extract: 1. Carrier Name, SCAC Code & Master BOL # 2. Shipper (Origin Address, Contact, Pickup Date) 3. Consignee (Destination Address, Delivery Window) 4. Third-Party Freight Bill-To Information 5. Commodity Breakdown Table: [Handling Units | Package Type | NMFC # | Class | Weight (LBS/KG) | Hazardous Material (Y/N) | Description of Articles] 6. Special Instructions & Temperature Control specs Sanitize output: Output markdown table and clean JSON object. Strip commas from numerical weights.

Analyze the uploaded hotel folio / lodging receipt. Split every daily charge into distinct accounting expense categories: - Room Rate (per night) - Occupancy Tax / City Tourism Tax - Resort / Amenity Fees - Food & Beverage / Room Service (itemized separate from lodging) - Parking / Valet - Incidentals (Laundry, Minibar) Provide output as a daily chronological ledger: [Date | Expense Category | Amount | Currency | Payment Method Card Last 4 Digits]. State grand total and confirm whether sub-charges equal folio total.

Extract structured billing information from this Medical Superbill / Health Claim form. Fields to extract: - Patient ID & Date of Birth (Mask DOB to YYYY-XX-XX for HIPAA compliance) - Rendering Provider Name, NPI Number & Taxonomy Code - Date of Service (DOS) - Primary & Secondary Diagnosis ICD-10 Codes - Procedure Line Items: Array of [CPT/HCPCS Code | Modifier | Units/Quantity | Billed Amount] - Total Charges & Co-Pay Collected Amount Never hallucinate diagnosis codes; return exact alphanumeric strings found in document.

Perform precision extraction of tax information from this scanned US IRS tax document (W-2 or 1099-NEC/MISC). Detect the document type automatically, then extract box-by-box: - Form Type & Tax Year - Payer/Employer Name, Address, and Federal EIN - Recipient/Employee Name, Masked SSN (***-**-1234) - Box 1 (Wages/Nonemployee Compensation) - Box 2 (Federal Income Tax Withheld) - Social Security / Medicare Wages & Tax Withheld (if W-2) - State & Local Tax breakdowns (Boxes 15-20) Format as strict key-value JSON matching IRS electronic filing schema.

Extract identity data from this passport or ID card scan. Task: 1. Decode the Machine Readable Zone (MRZ) if present (TD1, TD2, or TD3 format): verify checksums for document number, birth date, and expiry date. 2. Visual Zone Extraction: - Full Name (Surname, Given Names) - Document Number - Issuing Country / Nationality (ISO 3166-1 alpha-3 code) - Date of Birth (YYYY-MM-DD) - Expiration Date (YYYY-MM-DD) - Sex / Gender 3. Flag any visible signs of digital tampering or mismatched fonts between Visual Zone and MRZ.

You are a bilingual trade logistics specialist. Parse this Japanese (日本語) or Chinese (中文) commercial customs invoice into English structured data. Extract: - Exporter & Importer details (original text + romanized English) - Commercial Invoice Number & Date - Port of Loading (積出港) & Port of Discharge (陸揚港) - Incoterms (e.g. FOB, CIF, DDP) - Line Item Details in English: [Original Description | English Translated Description | HS Tariff Code (6-8 digits) | Qty | Unit Price | Currency | Total] - Total Value in original currency and declared USD equivalent if stated.

Process this low-contrast, crumpled, or partially faded thermal store receipt. Guidelines: - Reconstruct damaged text using contextual bounding boxes (e.g., if item says 'M_LK 2% GAL', infer 'MILK 2% GAL'). - Separate regular purchases from bottle deposits, state sales taxes, and tip amounts. - Extract: * Store / Merchant Name & Store # * Transaction Timestamp & Register # * Cashier ID * Itemized List with quantities and unit discounts applied * Payment Method (Cash, Visa, Amex, Apple Pay) and approval auth code * Confidence Score (0.00 to 1.00) for each line item.

Abstract the critical business and legal terms from this multi-page commercial lease agreement. Output a clean 1-page executive summary table: 1. Parties: Landlord legal entity, Tenant legal entity, Guarantor 2. Premises: Building name, Suite #, Rentable Square Footage (RSF) 3. Term: Commencement Date, Rent Commencement Date, Expiration Date 4. Base Rent Schedule: Annual & Monthly rate per square foot across all lease years 5. Operating Expenses: Triple Net (NNN) vs Full Service Gross, Tenant Pro-Rata Share % 6. Security Deposit & Letter of Credit Amount 7. Renewal & Expansion Options (notice deadlines and pricing terms) 8. Permitted Use & Exclusivity Clauses.

Reconcile this e-commerce settlement report (Amazon FBA bi-weekly statement or Shopify Payments export). Categorize every transaction into accounting buckets: - Gross Product Sales (Principal) - Shipping Credits & Gift Wrap Credits - Marketplace Facilitator Sales Tax Collected - Amazon Referral Fees / Commission - FBA Fulfillment Fees (Pick & Pack) - FBA Monthly Storage Fees & Long-term Storage Surcharges - Customer Refunds & Restocking Fees - Promotional Rebates & Coupons - Advertising Spend (Sponsored Products Deduction) - Net Bank Transfer Payout Amount. Check math: Sum of categories MUST equal net transfer payout.

Parse this complex cloud/SaaS billing invoice (e.g. AWS, Snowflake, Stripe, Datadog). Extract: - Platform / Vendor Name & Billing Account ID - Service Period (Start Date - End Date) - Fixed Recurring Seat Charges (Tier name, seat count, unit price) - Metered Consumption Line Items: [Service Name | Metered Metric (GB, API Calls, Compute Hours) | Units Consumed | Tier Rate | Subtotal] - Credits Applied (Free tier allowances, promotional credits) - Overage Penalties - Invoice Total & Next Scheduled Auto-Debit Date Format as JSON with distinct arrays for 'recurring_charges' and 'usage_based_charges'.

Analyze this end-of-day POS summary printout (Toast, Clover, Square Z-Report). Extract and balance the daily numbers: - Date, Shift Period, Terminal ID - Gross Food Sales, Beverage/Alcohol Sales, Merchandise Sales - Comps, Voids & Employee Discounts Total - Net Sales & Tax Collected Breakdown (Food Tax vs Liquor Tax) - Tender Breakdown: Cash, Visa, Mastercard, Amex, Gift Cards, Delivery Integrations (UberEats, DoorDash) - Gratuity: Auto-Gratuity, Credit Card Tips, Cash Tips - Cash Drawer Audit: Starting Float, Cash Sales, Paid Outs (Petty Cash), Expected Cash vs Actual Drop, Cash Over/Short (+/-). Highlight any discrepancy greater than $5.00 in bold red.

Extract the full loan payment amortization schedule from this promissory note or bank debt statement. Extract: 1. Core Terms: Principal Loan Amount, Annual Interest Rate, Term (Months), Compounding Frequency, Origination Date, First Payment Due Date. 2. Schedule Table: For each payment period, extract [Payment # | Due Date | Monthly Total Payment | Principal Component | Interest Component | Escrow/Insurance (if any) | Remaining Principal Balance]. 3. Calculate summary metrics: Total interest paid over lifetime of loan and balloon payment at maturity if applicable. Output direct CSV format ready to paste into Excel.

Parse this official university academic transcript scan. Extract: - Institution Name, Student Name, Student ID # - Degree Program, Major, Minor, Graduation Date / Standing - Chronological Semester Breakdown: * Term (e.g. Fall 2024) * Course Code & Course Title * Credits Attempted & Credits Earned * Grade Received * Grade Points * Term GPA - Cumulative Summary: Total Cumulative Credits Earned, Overall Cumulative GPA. Flag any repeated courses or pass/fail non-graded credit units.

Parse this property casualty insurance claim estimate (e.g. Xactimate output or auto repair damage report). Extract: - Claim Number, Policyholder Name, Date of Loss, Adjuster Name - Depreciation Schedule (Replacement Cost Value RCV vs Actual Cash Value ACV) - Deductible Amount - Room-by-Room / Area-by-Area Repair Line Items: [Area/Room | Line Description | Action (Remove / Replace / Clean) | Quantity | Unit | Unit Price | Labor Cost | Material Cost | Depreciation | Total Net Claim] - Grand Totals: Net Claim Amount Payable, Depreciation Holdback.

Extract comprehensive energy consumption metrics from this utility bill (Electric, Natural Gas, or Water). Fields required: - Utility Provider Name & Customer Account Number - Service Address & Meter Number - Billing Cycle Dates (From Date - To Date) & Number of Days in Cycle - Previous Meter Reading vs Current Meter Reading (Multiplier applied) - Consumption: Total kWh (or Therms/CCF) used - Peak Demand: kW peak demand and time of peak (if commercial) - Rate Tariff Schedule: Base Service Charge, Delivery/Transmission Charge, Supply/Generation Charge, Regulatory Environmental Surcharges - Total Amount Due & Due Date - Prior 12-Month Historical Average Usage (if chart/table present).

Extract all financial components from this employee pay stub / payroll journal (ADP, Paychex, Gusto). Structure the data into three balanced columns: 1. Earnings: [Earning Type (Regular, Overtime, Bonus, PTO) | Hours | Rate | Current Amount | YTD Amount] 2. Taxes Withheld: [Federal Income Tax, Social Security (OASDI), Medicare, State Withholding, Local/City Tax | Current | YTD] 3. Deductions: [Pre-Tax (401k, Health Insurance, HSA) | Post-Tax (Roth, Garnishments) | Current | YTD] 4. Net Pay Verification: Confirm Gross Earnings - Total Taxes - Total Deductions = Net Direct Deposit Amount.

Compare the line items in this uploaded Packing Slip against the associated Purchase Order. For each line item, evaluate: - Part / Item SKU - Ordered Quantity (PO) - Shipped / Received Quantity (Packing Slip) - Discrepancy Delta (+/- Shortage or Overage) - Backorder status - Damaged Goods note Generate a bold 'RECEIVING DISCREPANCY EXCEPTION REPORT' listing only the items where Received Quantity does not equal Ordered Quantity, including the estimated financial impact of missing goods.

Transform this messy credit card statement PDF into clean, categorized financial data. Rules: 1. Clean merchant names: Strip obscure billing codes (e.g., convert 'AMZN MKTP US*2K83J1 SEATTLE WA' to 'Amazon Marketplace'). 2. Assign a standard category to each row (e.g. Software, Office Supplies, Meals & Entertainment, Travel, Utilities). 3. Flag foreign transaction fees or recurring subscriptions. 4. Output columns: [Post Date | Transaction Date | Clean Merchant | Category | Amount | Card Member Name]. Ensure output is sanitized against Excel formula injection (prepend single quote if text starts with =, +, -, @).

Extract and organize the construction subcontractor bid estimate according to standard CSI MasterFormat divisions. Group into divisions (e.g. Div 03 Concrete, Div 06 Wood & Plastics, Div 22 Plumbing, Div 26 Electrical): - Contractor Name, License #, Bid Proposal Date - Scope of Work Included vs Excluded - Labor Hours & Hourly Burden Rate - Material Schedule: [Material Description | Qty | Unit | Unit Price | Total] - Equipment Rental Costs - Overhead & Profit (O&P) percentage markup - Total Base Bid & Voluntary Alternates.

Extract inspection parameters from this Manufacturing Quality Certificate / First Article Inspection Report (FAIR). Extract: - Part Name, Drawing Part #, Revision Letter, Batch/Lot # - Material Heat Number & Chemical Composition Certification - Dimensional Inspection Table: [Feature/Dimension # | Nominal Spec | Tolerance Low | Tolerance High | Measured Value | Pass/Fail Status] - Surface finish & hardness test results - Quality Inspector Stamp, Signature & Date Flag any dimension that is marked Fail or is within 5% of tolerance limits.