Everything you need to make sense of money

CSV exports from Chase, Wells Fargo, or Bank of America. Excel downloads from HSBC, Barclays, and Lloyds. PDF statements from Revolut, N26, Wise, Monzo, Starling, or hundreds of local and regional banks. TXT dumps from older systems. Scanned image PDFs from banks that still email you a print-ready file every month. All of it becomes structured, categorized rows in your Karchu workspace.

The parsing pipeline is deliberately layered. CSV and XLSX files are parsed directly in your browser using a schema-detection stage that identifies which columns hold the date, description, debit, credit, and running balance. Because the parse happens client-side, raw file contents never leave your device unless you deliberately save the extracted data. PDFs try text extraction first (fast and lossless for digital statements) and fall back to on-device OCR for scanned images. Multi-page statements are recomposed automatically, wrapped descriptions are joined, and rows split across subtotal boundaries are recombined into a single record.

A rule is a small pattern (a merchant name, description keyword, amount range, or account type) that tags any matching transaction with a category. Two months from now, when you wonder why a charge landed in Subscriptions instead of Software, you can open the rule that decided it and edit it. The change applies retroactively to every past transaction that matched and to every future one that ever will.

The engine ships with sensible defaults for hundreds of common merchants: Uber, Lyft, Amazon, Netflix, Spotify, Whole Foods, Starbucks, Airbnb, Vercel, GitHub, Stripe, and more. Rules chain and stack, so a specific catch (AMAZON PRIME → Subscriptions) sits above a general one (AMAZON → Shopping), and the more specific rule wins. Amount-range rules split ambiguous merchants automatically without hand-labeling anything.

The default view shows income versus expenses over your chosen date range, current cash-flow trend, the top ten merchants by spend, a category breakdown, and a leaderboard of categories that grew or shrunk fastest versus the prior period. Drill into any category, filter by account, filter by merchant, or slice by tag.

Small businesses get an extra view surfacing cash-in versus cash-out per month, top expense categories, and unusual-spending flags. Freelancers get a project-tagging model where transactions carry a project tag on top of the category, so a single trip to Home Depot can be split across the correct client engagement without breaking category totals. Every dashboard exports as PNG for reports or CSV for downstream analysis.

Drop a JPG, PNG, or PDF into the receipts view and Karchu extracts the vendor, date, subtotal, tax, tip, total, and every line item. Receipts are matched to their transaction so your books show the full detail without any manual data entry. This matters most for reimbursement approvals, tax deductions, client-billable expenses that need itemization, or splitting a joint dinner into two categories cleanly.

Owner (full control including billing), manager (all operational access), accountant (read plus category and rule editing), analyst (read-only across all data), employee (create receipts and see their own transactions), and viewer (read-only against a scoped subset). New members join by invitation only, invitations expire, and role assignments are enforced on the server, not in the UI. A modified client cannot escalate its own permissions. Audit logs record who invited whom, who changed which rule, and who exported which report.

Spin up as many organizations as you need under one login. Switching between them is one click; each carries its own members, rules, categories, and file vault. Nothing crosses over unless you explicitly copy a rule template between organizations.

Files live in a private vault. Encryption is applied in transit and at rest through our managed cloud infrastructure. We do not sell your data, we do not train third-party AI models on it, and we do not use bank-aggregation services that would require your online-banking credentials.

Character encoding is UTF-8 so foreign merchant names survive the trip. The file drops straight into QuickBooks, Xero, Wave, a Google Sheet, or your accountant's template without a middleware step. No per-export fees, no monthly export caps, and no premium tier required to get your own data back.