Synthetic demo data · Read-only · Run the full agent with approvals locally: npx closeloop-mcp
← Exception queue

Suggested match

Probable match awaiting approval (open), ₹78,504.53. Rule stage2.fuzzy_suggest; 2 record(s).
₹78,504.53Structuralopen12 Oct 2026bank_line bnk_000146

Explanation

Generated from the read-only evidence below. It never changes a match or a record.

LLM · openai/gpt-oss-120b

The bank transaction’s narration contains a truncated UTR that only partially matches the settlement’s full UTR.

The narration shows "KKBKN2628583" while the settlement record has the full UTR "KKBKN26285836939". The amount (₹78,504.53) and date (2026-10-12) are identical, giving the engine a confidence of 0.75. Because the token is marked as truncated, the match is suggested but not certain.

Recommended action

Verify whether the truncated UTR in the narration is a typo; if confirmed, approve the suggested match, otherwise reject it.

Requires human reviewConfidence 75%· Cites 2 evidence ids, all verified to be in the bundle

Evidence records

Read-only source records involved in this exception.

Bank statement linebnk_000146
txn date
2026-10-12
narration
NEFT/KKBKN2628583/PGW SETTL/GATEWAY SETTLEMENTS
credit
₹78,504.53
debit
₹0.00
balance
₹1,29,33,565.48
Settlementsetl_oTsP4buNJn7XAv
utr
KKBKN26285836939
amount
₹78,504.53
settled at
2026-10-12
created at
2026-10-12T09:03:00+05:30
item count
25
fee total
₹1,804.50
gst total
₹324.81

Rule trace

The deterministic rule that fired and what it saw.

{
  "rule": "stage2.fuzzy_suggest",
  "reason": "truncated_utr",
  "best": {
    "settlement_id": "setl_oTsP4buNJn7XAv",
    "utr": "KKBKN26285836939",
    "token": "KKBKN2628583",
    "token_source": "narration",
    "distance": null,
    "truncated": true,
    "confidence": 0.75,
    "amount_equal": true
  },
  "runner_up": null,
  "margin": 0.75,
  "suggested": {
    "settlement_id": "setl_oTsP4buNJn7XAv",
    "utr": "KKBKN26285836939",
    "token": "KKBKN2628583",
    "token_source": "narration",
    "distance": null,
    "truncated": true,
    "confidence": 0.75,
    "amount_equal": true
  },
  "candidates": [
    {
      "settlement_id": "setl_oTsP4buNJn7XAv",
      "utr": "KKBKN26285836939",
      "token": "KKBKN2628583",
      "token_source": "narration",
      "distance": null,
      "truncated": true,
      "confidence": 0.75,
      "amount_equal": true
    }
  ]
}

Audit history

Append-only. Actor: system, llm or human.

  1. llminvestigation

    source=llm · model=openai/gpt-oss-120b · review_policy_applied=false

  2. systemjournal draft

    applicable=false · reason=SUGGESTED_MATCH needs a matching decision, not a journal entry