Bank reconciliation automation that clears the matches and shows you the exceptions

Pulling in transactions and matching the routine

The mechanical heart of reconciliation is comparison: this payment on the statement is that invoice in the ledger, this receipt is that customer settlement. We pull transactions from your bank, through a bank feed, an API, or a statement import, depending on what your bank supports, and match them against the entries in your accounting system by amount, date, reference, and counterparty. The clean matches, which are most of them, clear automatically.

Matching is not always one-to-one, and a rigid rule breaks on the real world: a single bank deposit that covers three invoices, a payment that arrives a few days after the ledger date, a reference that is close but not exact. We combine deterministic rules for the clear cases with fuzzy and AI-assisted matching for the messy ones, so a payee name that is abbreviated or a rounded amount still finds its partner. Where the match is plausible but not certain, it is proposed for a person to confirm rather than booked silently.

• Pull transactions via bank feed, API, or statement import, whatever your bank supports
• Match on amount, date, reference, and counterparty for the clear cases
• Fuzzy and AI-assisted matching for abbreviated names, rounded amounts, and timing gaps
• One-to-many matching for a single deposit covering several invoices

Surfacing exceptions instead of burying them

The value of reconciliation is not in the matches that agree; it is in the ones that do not. A transaction on the bank statement with no entry in the ledger, a ledger entry with no bank movement, a duplicate, an amount that is close but wrong: these are where errors, missed postings, and occasionally fraud live. When they are hand-checked among hundreds of routine lines, they are easy to miss. When the automation clears the routine, they are all that is left, and they are impossible to miss.

We present exceptions as a clean, prioritised worklist rather than a raw mismatch dump. Each one carries the context a person needs to resolve it: the bank line, the candidate ledger entries, why the automation could not match them, and how old the discrepancy is. Your finance team spends its reconciliation time investigating real problems with the information already gathered, instead of ticking off matches a machine could have cleared.

• Unmatched bank transactions with no corresponding ledger entry
• Ledger entries with no matching bank movement
• Duplicate transactions and near-miss amounts
• Each exception carries its context and age, ready to investigate

Human judgement on the writes that matter

Clearing a confident, exact match is safe to do automatically, and that is where most of the time savings come from. But creating a new ledger entry to resolve an exception, writing off a small difference, or reclassifying a transaction is a decision, and decisions stay with a person. The automation prepares the proposed entry and the reasoning; a human approves it before anything is written to the ledger.

You control where the line between automatic and reviewed sits, and we make it explicit. High-confidence matches within your tolerances can clear on their own; anything below that confidence, above an amount you set, or involving a write-off waits for approval. This is the human-in-the-loop principle applied to reconciliation: the machine does the exhausting comparison, the person makes the calls that affect the books.

A full audit trail, and honest scope

Every match, whether cleared automatically or approved by a person, is logged: which bank line met which ledger entry, on what basis, and who confirmed it if a human was involved. Unresolved exceptions are tracked with their age so nothing quietly lingers. When your auditor asks how a period was reconciled, the evidence is complete and reconstructable, which is exactly what reconciliation is supposed to provide in the first place.

We build the matching, the exception handling, and the trail, running on infrastructure you own and connected to your bank and accounting system through supported interfaces. We are honest about the boundary: reconciliation automation makes the process fast and clean and puts the right things in front of your team, but the judgement on exceptions, the sign-off on the period, and any accounting treatment stay with your finance professionals. We clear the noise so their expertise lands where it counts.