Picture a Monday morning initial consultation. Your client arrives with a folder: two years of T4 slips, two Notices of Assessment from CRA, three months of pay stubs, and a set of corporate financial statements from a side business. Before you can discuss support ranges, custody options, or next steps, someone needs to pull the numbers out of those documents and enter them into your calculation -- correctly.
That means finding Line 15000 on the NOA (total income, not Line 23600 net income -- a common mix-up that can swing guideline income by tens of thousands of dollars). It means distinguishing T4 Box 14 employment income from T4A pension income reported gross, which clients routinely quote as the net deposit they actually receive. It means cross-referencing pay stub YTD totals against the T4 to catch mid-year job changes. For a two-party file, you are looking at 40 to 80 individual data points, each one a potential transposition error that flows directly into your support calculation.
This is the task that consumes the first 20 to 30 minutes of every initial consultation and recurs every time new financial disclosure lands on your desk. Industry data shows lawyers spend 28 to 48 percent of their working time on non-billable administrative work. For family law practitioners, financial data entry is one of the largest single contributors to that number.
The Problem We're Solving
Family law professionals spend a significant amount of time on data entry. A typical support calculation requires income information from multiple sources — T4 slips, T1 General returns, CRA Notices of Assessment, corporate financial statements, and pay stubs. Manually transcribing figures from these documents into a calculator is tedious, time-consuming, and prone to error.
For a straightforward matter with two employed parties, you might need to enter dozens of data points. For complex cases involving self-employment, corporate income, or multiple income sources, the data entry burden multiplies. Every mistyped number is a potential error in the final support calculation.
How It Works
The extraction feature works in three straightforward steps:
1. Upload Your Documents
From within any child support or spousal support calculation, click the "Upload Documents" button. You can upload PDF files, scanned images, or photographs of financial documents. The system accepts T4 slips, T4A slips, T5 slips, CRA Notices of Assessment, T1 General returns, pay stubs from major Canadian payroll providers, and corporate financial statements.
2. AI Extracts the Data
Our AI model analyzes each uploaded document, identifies the document type, and extracts the relevant financial fields. For a T4 slip, this includes employment income, CPP contributions, EI premiums, income tax deducted, and other reportable amounts. For a Notice of Assessment, the system pulls total income, net income, taxable income, and any reassessment adjustments.
Calculate child support, spousal support, and property division in minutes.
The extraction uses a combination of document layout analysis and natural language understanding. It can handle standard CRA forms, employer-generated T4s in various formats, and financial statements with different layouts.
3. Review and Confirm
Extracted data is presented for your review before being applied to the calculation. Each extracted field shows the source value from the document alongside where it will be entered in the calculation. You can accept all extracted values, modify individual fields, or reject the extraction entirely. Nothing is applied automatically without your confirmation.
This review step is critical. While the extraction handles most standard documents well, financial documents vary widely in format and quality, and professional responsibility demands that lawyers verify the data underlying their calculations. The AI is a tool to accelerate your workflow, not a replacement for professional judgment.
Supported Document Types
- T4 — Statement of Remuneration Paid: Employment income, taxable benefits, CPP/EI contributions, union dues, pension adjustments
- T4A — Statement of Pension and Other Income: Pension income, annuities, self-employed commissions, RESP payments
- T5 — Statement of Investment Income: Dividends (eligible and other-than-eligible), interest income, capital gains dividends
- CRA Notice of Assessment: Total income, net income, taxable income, refund or balance owing, RRSP deduction limit
- Pay stubs: Gross pay, deductions, year-to-date totals (supports major Canadian payroll formats)
- Corporate financial statements: Revenue, expenses, net income, shareholder compensation, retained earnings
Accuracy and Privacy
Extraction accuracy depends on document quality and format. Clean, electronically generated PDFs (such as those downloaded from CRA My Account) perform best. Scanned documents and photographs may require more review, though the system handles most legible documents well. We recommend always reviewing extracted values against the source document before confirming.
Privacy is paramount when handling financial documents. All uploaded documents are processed using encrypted connections and are not retained after extraction is complete. We do not use uploaded documents to train our AI models. Our data handling practices comply with Canadian privacy legislation including PIPEDA, and our infrastructure is hosted in Canada.
Impact on Workflow
The feature is particularly valuable for initial consultations, where you need to produce a preliminary support estimate quickly based on documents the client brings to the meeting. Upload the documents, review the extracted data, and generate a calculation in minutes rather than spending the first half of a consultation on data entry.
For ongoing files, extraction saves time each time new financial disclosure arrives. When the other party produces updated T4s or a new Notice of Assessment, you can upload, extract, review, and update your calculation in a fraction of the time it would take to enter everything manually.
Common Data Entry Errors This Prevents
Every family law practitioner has caught -- or missed -- at least one of these:
- Line 15000 vs. Line 23600: Using net income instead of total income from a Notice of Assessment. The difference can be $5,000 to $20,000 depending on RRSP contributions and other deductions, directly changing guideline income and table amounts.
- T4 Box 14 vs. total income: T4 employment income excludes rental, investment, and self-employment income. Entering Box 14 when the calculator expects total income understates the support obligation.
- Gross vs. net pension: T4A pension income is reported gross. Clients often quote the net deposit. The difference affects both the income determination and the tax calculation.
- Transposed digits: $85,400 entered as $84,500 is a $900 error. In a child support table lookup, that can shift the monthly amount by $20-50. Across a multi-year order, it compounds.
- Stale YTD figures: Using a pay stub from March to annualize income when the client changed jobs in April. The annualized figure is based on 3 months at the old salary, not 12 months at the new one.
AI extraction does not eliminate the need to review -- but it eliminates the transcription step where most of these errors originate. The extracted values are presented side-by-side with source documents, making verification faster and more reliable than manual comparison.
Getting Started
AI data extraction is available now for all Divorcepath Pro subscribers at no additional cost. Simply open any support calculation and look for the "Upload Documents" option in the income section. You can also access the feature from the document management area of any client file.
We'll continue refining the extraction models based on user feedback and expanding the range of supported document types. If you encounter a document type that isn't well supported, let us know through the feedback button that appears after each extraction.
Try it today at divorcepath.com.
