Oracle NetSuite and Plaid integration

What moves between them

Transaction and balance data flows from Plaid into Oracle NetSuite daily or on demand. When a bank transaction clears in Plaid, ml-connector reads it and matches it against vendor payments and expense records in NetSuite, then posts a reconciliation entry into the appropriate cash GL account. Vendor bank account details move bidirectionally: bank account changes in NetSuite vendor records are validated against accounts linked in Plaid to catch mismatches before payment. NetSuite does not write back to Plaid; the bank connection is read-primary.

How ml-connector handles it

ml-connector stores Plaid API credentials and per-user access tokens encrypted in the database, and refreshes user tokens via the Plaid Link flow when they expire. It validates webhook signatures from Plaid using the ES256 JWT algorithm and public key lookup, rejecting any signature with a timestamp more than 5 minutes old. For NetSuite, it uses OAuth 2.0 bearer tokens, refreshing when a call returns 401, and accepts the 60-minute token lifetime as a scaling constraint for batch reconciliation runs. Transaction matching is rule-based: Plaid transaction date and amount are matched against NetSuite vendor payment date and bill amount within a configurable tolerance window, and mismatches generate an alert for manual review. Bank account validation compares the last 4 digits and institution routing number held in NetSuite against Plaid's account data. Rate limits are honored with exponential backoff and jitter. Every transaction move and validation carries a full audit trail.

A real-world example

A mid-sized professional services firm uses Oracle NetSuite to manage vendor payments and expense reimbursements, and maintains operating accounts at three banks. Before the integration, the accounting team imported daily bank statements manually, matched them to paid vendor invoices one by one, and investigated discrepancies that arose from timing differences or data entry errors. With Oracle NetSuite and Plaid connected, each day's cleared transactions flow in automatically, vendor payments are matched and flagged for review only when amounts or dates fall outside tolerance, and a single dashboard shows the firm's cash position across all three banks. Month-end reconciliation now takes one day instead of three, and the finance team catches payment failures and duplicate payments faster.

What you can do

• •Sync cleared bank transactions from Plaid into Oracle NetSuite daily to reconcile cash and GL accounts.
• •Match vendor payments in NetSuite against cleared bank transactions in Plaid, flagging mismatches for review.
• •Validate vendor bank account details against Plaid's connected accounts before initiating payments.
• •Authenticate with Plaid API keys and per-user access tokens, refreshing when needed, and NetSuite OAuth bearer tokens with automatic 401 retry.
• •Track every transaction move, validation, and match in a complete audit trail with timestamp and user context.

Questions

Which direction does data move between Oracle NetSuite and Plaid?

Transaction and balance data flows primarily from Plaid into Oracle NetSuite for reconciliation and validation. Vendor bank account details move bidirectionally for validation purposes. Oracle NetSuite does not write back to Plaid; the Plaid connection is read-primary for financial data.

How does ml-connector handle Plaid webhook signatures and token expiry?

ml-connector validates every webhook from Plaid using ES256 JWT signature verification with public key lookup, and rejects any request with a timestamp older than 5 minutes. Per-user access tokens are refreshed via the Plaid Link flow when they expire, and the vendor account linked to that token is notified of any link failures.

How are vendor payments matched between the two systems?

ml-connector compares Plaid transaction date and amount against Oracle NetSuite vendor payment records within a configurable tolerance window (typically 2-3 days and a small dollar variance). Matches that fall outside tolerance are flagged as exceptions for manual review, and the full match logic and results are logged in the audit trail.