About
LedgerBug is smart, self-serve bookkeeping for people who aren’t accountants: freelancers, new LLCs, solo founders, and side-hustlers who need clean books but will never open QuickBooks. Connect your bank and the AI does the bookkeeping; you stay in approval.
- Connect your bank. Transactions flow in automatically through Plaid.
- AI categorizes everything. A six-step pipeline (exact-match → fuzzy → pattern → Claude → review queue → learns from your corrections) categorizes each transaction and never makes the same mistake twice.
- Review only what needs you. A Needs Attention feed surfaces the handful of real issues (bank-vs-book gaps, uncategorized items, overdue invoices and bills, duplicates) with exact numbers, not noise.
- Ask LedgerBug. A floating AI assistant answers plain-English questions (“how much cash do I have?”), teaches the concepts (cash vs. accrual, when to make a journal entry), and proposes the actual entry as a one-click Approve & post card: journal entries, invoices, bills, recategorizations.
- See your financials, and hand off to a CPA. P&L, Balance Sheet, Trial Balance and GL; send invoices and track bills; then invite your accountant in, or share a no-login, read-only package.
Under the friendly surface it runs a real double-entry ledger, so you never have to think in debits and credits unless you want to. Multi-business is first-class: one owner can track several ventures, with “Activities” for sub-ventures inside a business.
LedgerBug vs. LedgerClaw: LedgerClaw is the firm OS for accountants managing many clients’ books; LedgerBug is the front office for the business owner doing their own, with a clean path to hand off to a CPA (who might be running LedgerClaw on the other side). Same GL-engine DNA, opposite audience.
It’s live today in invite-only access at ledgerbug.app, behind a waitlist.
The onboarding
The whole pitch is in the first minute. You list the businesses you run: a freelance gig, a rental, a side shop, and each one gets its own books. Pick the type of work and LedgerBug tunes the starting chart of accounts for you, then sets the entity structure. Review, and it builds the books for every venture at once, keeping each one’s transactions separate while still showing your combined cash.
How I built it
LedgerBug was forked from the LedgerClaw codebase and surgically stripped down to core bookkeeping: squashing a 32-migration history into one clean baseline and cutting everything that wasn’t essential (firm management, roles, time tracking) to get a focused, consumer-first product on its own infrastructure.
- Real double-entry under a friendly surface. Every action resolves to balanced journal entries; posted entries are immutable (void-only) and audit-logged. The AI proposes, the user approves, so an LLM never silently writes to the ledger.
- Agentic AI with guardrails. Claude (Sonnet) with tool-use over the live books; every tool is scoped to the user’s own businesses, and write actions only ever produce a draft the user confirms.
