The attorney has the documents.
The case builds itself.
Sarah's new client walked in with a folder. Paystubs, last year's tax return, a credit report he pulled, two bank statements. No portal, no waiting. She opens her app, makes a case, and drops the documents in one at a time. Six minutes later she has a Chapter 7 petition with the schedules built, the means-test computed, and every creditor address verified against the credit report.
New case
Top bar, + New → New Case. Bankruptcy practice, individual debtor. The wizard opens with a "Start from intake" card on top — Sarah's not using it this run, so she scrolls past and works through the stepper instead.
Petition step
Chapter 7, individual debtor, district of choice. Saving here creates the case and the first draft filing in one move, with the correct B 101 federal voluntary petition seated automatically.
Pull data from anywhere
Data tab. Pull data button opens a menu of every source the firm has configured: credit bureaus, paystubs, mortgage statements, tax returns, bank statements, CSV import. Each source is a recipe — what fields it can populate, where they land in the schema. Sarah picks Paystub.
Paystub extracted, ready to confirm
She drops the PDF. The extractor runs and shows the fields it found — employer, position, hire date, gross monthly, net monthly, pay frequency. Each with a confidence score. Sarah glances over the list, hits Confirm. Every field lands in the schema as one entry, sourced to the paystub, timestamped.
Two sources, one field
Sarah jots down the gross-monthly the client mentioned over the phone — $5,800. The paystub she just dropped in says $4,200. The field flags the disagreement the moment the second write lands. No silent overwrite, no buried discrepancy. Every source stays on the record with its value next to its name.
The credit report builds the schedule
Pull data → Credit report. The extracted trade-lines arrive as a list: balance, account number, monthly payment, address, type. Confirm the batch, and creditors[] now has four rows that map straight into Schedule E/F. No re-keying. Account numbers are verified to four digits, addresses are normalized.
![Credit report extraction adding four creditor rows to creditors[]](screenshots/ch7-05-credit-creditors.png)
Tax return and bank statement
Same flow, twice more. The tax return drops in last year's W-2 wages, refund amounts, dependents. The bank statement adds recurring expense categorization. Every entry shows up on the same timeline, every value carries its source.
Means test computes itself
Overview → Means Test sub-tab. Income lines are pre-filled from the paystub and tax return. The presumption-of-abuse calculation runs live: not above median. Chapter 7 is on the table.
Why the number is the number
Sarah taps the annualized-income cell on B 122A-1. The rail shows the formula — gross_wages × 12 — and below it, every input with its currently-resolved value and where that value came from. Output at the bottom: the cell's actual filled number. No spreadsheets, no "trust me, the law firm's macro spreadsheet does this," no opaque math. The case file can answer "why is this $50,400?" without a phone call.
The schedules are built
Forms tab → Filings. The Ch 7 petition group is there: B 101, B 106 Sum, Schedule A/B, D, E/F, G, H, I, J, Statement of Financial Affairs. Each is populated from the schema. Yellow markers show what's still missing — typically signature blocks and a couple of attestations.
Petition ready to file
Select the Chapter 7 group. The full schedule tree expands — voluntary petition, schedules A/B, C, D, E/F, G, H, I, J, the Statement of Financial Affairs, the means-test forms. Each leaf is populated from the schema; the rail shows readiness state across the whole packet. Download the merged PDF, sign, file. Six minutes from new-case to court-ready — and Sarah didn't type a single line of debtor data; she pulled four documents.
Loose document? Drop it on the assistant.
Sarah's client emails over a stray document she didn't ask for — a paystub from a side gig. Instead of figuring out which Pull-Data flow to run, she opens the assistant on the right and drags the file in. The assistant identifies the document type, extracts the relevant fields, and offers to write them to the schema. Same approval gate, same source tagging — just a friendlier entry point.
The other direction.
Same case, same outcome — but starting from the client and the portal. Watch the trust moment when the debtor's typed income meets what the paystub says.