Data ingestion

PACIN scores records that already have activity. You send that activity in. Then you request a score.

API ingest

POST /v1/ingest accepts a batch of transactions. Use a unique idempotency key so a retry does not double-write. Live data needs a valid consent_id.

curl -X POST https://app.pacinnetwork.com/api/v1/ingest \
  -H "Authorization: Bearer $PACIN_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: ingest-2026-09-14-001" \
  -d '{
    "source": "json",
    "rail": "mpesa",
    "country_code": "KE",
    "consent_id": "con_01JXXXXXXXXXXXXXXXXXXX",
    "records": [
      {
        "entity_id": "pac_KE_0001",
        "transaction_type": "credit",
        "amount": 25000,
        "currency": "KES",
        "timestamp_utc": "2026-05-13T08:23:14Z",
        "source_record_id": "txn-001"
      }
    ]
  }'

Each record needs an entity id, a type (for example credit or debit), amount, currency, timestamp, and a source id from your system.

File upload

For a one-off history file, use the bulk upload tool in your institution account. Supply: entity identifier, date, amount, transaction type, rail, and currency.

Consent

Production ingest without consent is rejected. Sandbox keys may use test entity IDs for development. See Consent.