A reliability layer that turns messy bank and invoice files into validated structured data for software and automation workflows — with a confidence score and full provenance on every field.
Deterministic parsing you can trust with financial data — every value scored, traced, and checked.
Structural, type and balance checks on every file. Each field gets a heuristic confidence score, never dressed up as certainty.
Every output value carries where it came from: source row, parser, mapping, and every transformation applied along the way.
Fields the parser isn't sure about are routed to a human review queue instead of silently guessed. You stay in control of the edge cases.
Wrong encodings, mixed date formats, broken headers, ambiguous columns — recovered into a clean canonical ledger without dropping rows.
Four honest steps. Nothing is inferred that can't be shown.
Drop a bank CSV, invoice, or text file. Type is sniffed from content, not the extension. Files stay on your server.
Deterministic parsers map to a canonical schema, run checks, and score each field's confidence.
Uncertain fields surface in a review queue. Confirm or correct them; duplicates are flagged, never deleted.
Pull clean CSV or JSON through the API or dashboard, with provenance attached for your audit trail.
One endpoint per tool. One envelope for every response — success or rejection.
curl https://apodexa.net/api/v1/normalize_bank_csv \ -H "Authorization: Bearer $APODEXA_KEY" \ -F "file=@bank_202405.csv" \ -F "locale=ja-JP" \ -F "currency=JPY"
{
"data": { "rows": 2432 },
"validation": {
"passed": true,
"checks": [ "balance_reconciles", … ]
},
"confidence": { "transaction_date": 0.99 },
"uncertain_fields": [],
"provenance": { "parser": "normalize_bank" },
"error": null
}
Amounts are decimal strings plus integer minor units — never floats. Full reference at the dashboard API tab.
Sign in to process your first file and see every field scored and traced.
Open the dashboard