Everything you need to know about LedgerBin and AI-powered transaction categorization.
Upload a bank statement CSV and LedgerBin's AI reads each transaction description, amount, and date pattern. It matches transactions to your specific Chart of Accounts — not a generic category list. The AI uses Claude to understand context (e.g., 'AMZN MKTP US' maps to Office Supplies if that's how your client categorizes Amazon purchases). Accuracy starts around 85% and improves to 95%+ as you make corrections.
LedgerBin accepts CSV files from any bank. The AI automatically detects column mappings — date, description, amount, debit/credit — regardless of column names or ordering. You don't need to reformat your CSV before uploading. We support date formats from US, UK, EU, and ISO standards.
LedgerBin exports categorized transactions as QBO-compatible CSV and IIF files that you upload directly to QuickBooks Online. This gives you full control to review categories before they hit the client's books. We're building direct QBO API sync for later in 2026.
Yes. LedgerBin exports Xero-compatible CSV files with the correct column format for Xero's bank statement import. Like QBO, you review everything in LedgerBin first, then upload the clean file to Xero.
When you correct a categorization, LedgerBin stores that correction tied to the transaction pattern and client. Next time a similar transaction appears for that client, the AI uses your correction as a strong signal. Over weeks of use, the AI learns each client's specific categorization preferences.
Absolutely — that's the whole point. You upload or enter your client's Chart of Accounts, and the AI categorizes against those specific accounts. No generic categories forced on you. Each client can have a completely different COA.
The free trial gives you 1 client, 100 transactions, and 14 days. No credit card required. That's enough to upload a real bank statement and see how the AI handles your client's transactions. After the trial, plans start at $29/month.
Bank feed rules in QBO/Xero are string-matching: if description contains 'Staples', categorize as Office Supplies. They break when descriptions vary ('STAPLES #1234' vs 'STP*STAPLES.COM'). LedgerBin's AI understands context and variations without manual rule creation. It also works with CSV imports, not just bank feeds.
LedgerBin runs on Supabase with row-level security. Each firm's data is isolated at the database level. We use TLS encryption in transit and AES-256 at rest. Transaction data is processed by Claude AI (Anthropic) with enterprise data handling — your data is never used to train models.
It depends on your plan. Starter supports up to 5 clients, Professional up to 25, and Agency is unlimited. All plans include unlimited transactions per client. See our pricing page for full details.
Start a free trial and see LedgerBin in action, or check our pricing.