Status: 0.0-sketch. A goal on the record, not yet a spec.
Saybooks owns operational truth and derives balanced journal lines from it (core_journal). The LEDGER OF RECORD — QuickBooks Online, Xero, an accountant's own system — owns the chart of accounts, manual journals, adjustments, and the close. The bridge is how the first feeds the second. We never compete with the ledger; we make it better fed than it has ever been.
What it is
Account mapping — our fixed derivation accounts (Accounts Receivable, Sales Revenue, Service Revenue, Sales Tax Payable, Cash, Customer Deposits, Customer Credits, Sales Returns & Allowances, Bad Debt Expense; later the p2p set: Inventory, GRNI, AP, COGS, price variance) mapped onto the customer's REAL chart. The mapping is workspace config, set once with the accountant, logged like everything else.
Export formats — journal entries in the shapes the targets actually import (Xero manual-journal CSV; QBO journal import; plain CSV for everyone else). Exact field layouts to be pinned against live trials, not documentation.
Hand-over tracking — what has been exported, through when. The journal itself stays a pure derivation (re-derivation is the truth); the bridge records the hand-over so month-end is a diff against last export, not a re-key and not a duplicate import. Late-arriving or voided facts after an export produce a visible correction entry in the next export, never a silent rewrite of the past.
What it is not
No chart of accounts of our own. No manual journals. No close. No bank feeds — the CSV import on unapplied cash is the honest boundary, and bank-feed integrations remain the moat we say out loud we have not crossed.
Acceptance test (the plan)
Xero demo company: free, resettable, imports manual journals — the primary target. Export a month of the demo fixture's books, import, verify the trial balance ties to core_journal's totals and AR ties to o2c_ar_aging.
QBO trial account: same exercise against QBO's journal import; document plan-level quirks honestly (import availability differs by plan/region).
The acceptance bar: an accountant who has never seen Saybooks can take one export file and land it in their ledger without asking us anything.
---
Change log: 0.0-sketch (2026-08-24) — recorded as a goal after the o2c 0.2 journal work, so the "operating layer in front of the books" positioning has its roadmap on paper.