Solutions Reconciliation

Automatic reconciliation.

Banks, acquirers, PSPs, and the ledger matched by rule. Whatever does not agree becomes an exception with an owner and a history, until it is resolved.

What is included

From import to the audit trail.

  • Sources

    Banks, acquirers, PSPs, ERPs, and the ledger itself, each in its own format: CSV, JSON, XML, camt.053, CNAB 240 and 400, OFX, and acquirer layouts. Direct extraction from databases with the Fetcher engine built in. Field mapping, validation, and duplicate detection on the way in.

  • Reconciliation rules

    Exact, with an amount tolerance, with a date lag, or by reference similarity. Evaluated in priority order; whatever is left over becomes an exception.

  • 1:N and N:M relationships

    One payment against several invoices, several Pix against one payout, the sum of TEDs against a single open receivable. The relationship is set by context, not forced into 1:1.

  • Confidence scoring

    Every proposed match gets a score from 0 to 100 across amount, currency, date, and reference. Above the threshold the match is confirmed automatically; below it, a person reviews it.

  • Fees and currencies

    Expected acquirer, gateway, or bank fees modelled before the comparison, so an expected difference does not become an exception. Cross-currency comparison on a base amount.

  • Exceptions

    Every discrepancy becomes an exception with an owner, a severity, and a history, resolved by manual confirmation, an accounting adjustment, or routing to Jira, ServiceNow, or a webhook.

  • Reporting

    A reconciliation summary, match detail, open items, exception aging, and fee variances, with export and a dashboard by context and period.

  • Audit trail

    Every match, exception, and resolution in an immutable log, hash-chained and integrity-checked. Operators' personal data is kept separately, pseudonymized.

Matcher

It works on its own or alongside the ledger.

Matcher is Lerian's reconciliation engine. It compares two sets of data from any origin and shows what agrees, what does not, and why. It runs on your own infrastructure or on Lerian Cloud; Midaz is one more source, not a prerequisite.

  • Standalone or integrated
  • Your infrastructure or Lerian Cloud
  • Immutable audit

For developers

  • Idempotent REST API

    Every write accepts an idempotency key; bearer token authentication, with the tenant taken from the JWT and never from the payload.

  • Native MCP server

    AI assistants operate Matcher with the permissions of whoever connects: contexts, runs, exceptions, and reports, over Streamable HTTP.

  • Webhooks and callbacks

    Exceptions delivered by webhook signed with HMAC-SHA256; the resolution comes back by callback from Jira, ServiceNow, or your own system.

  • Simulate before applying

    Test a rule or a fee table against the real data in the context, without writing anything.

  • Rule suggestions, approved by a person

    AI proposes rules from aggregates of the history; no raw data leaves the environment, and nothing becomes a rule without approval.

  • Go, PostgreSQL, Helm

    A modular monolith of seven modules, a database per tenant, and an official Helm chart for Kubernetes.

Contact

Leave your details and Lerian will get in touch to work out the best way to serve your operation.

Talk to Lerian

Or explore first: Sandbox Documentation GitHub