Sage 100 and Plaid integration

What moves between them

Bank transactions and balances flow from Plaid into Sage 100 for GL reconciliation. Transactions are fetched via Plaid REST endpoints and mapped to Sage 100 GL accounts based on merchant name, transaction type, or customer-configured mapping rules. Approved payments in Sage 100 flow out to Plaid, which initiates ACH transfers to vendor bank accounts. Plaid webhooks notify Sage 100 when transfers succeed or fail, updating the payment status in Sage 100's AP module. GL reconciliation feeds happen hourly; payment initiation is on-demand when an AP user approves a check or batch in Sage 100.

How ml-connector handles it

ml-connector stores Sage 100 credentials (username, password, company code, and SOAP endpoint URL) encrypted, and Plaid credentials (client_id, secret, and access token from Plaid Link) the same way. It polls Sage 100 GL accounts and bank transaction stubs hourly, and fetches full transaction detail from Plaid using REST pagination (cursor-based for transaction syncs). Each Plaid transaction is mapped to a Sage 100 GL account via matching rules configured per customer, then inserted as a journal entry if not already present. For payments, ml-connector accepts a list of Sage 100 AP invoices or manual checks approved for payment, creates an ACH transfer request in Plaid with an idempotency key tied to the check number, and polls Plaid's transfer endpoint until the transfer succeeds or fails. Plaid webhooks are registered to notify ml-connector of transaction sync availability and transfer status; each webhook is verified by decoding the JWT, checking the alg claim equals ES256, validating the iat claim is within 5 minutes, and comparing the request body SHA-256 to the claim. If a Plaid API call returns a 429 rate-limit error, ml-connector backs off with exponential jitter and retries. Every transaction and transfer carries an audit trail with timestamps, user, and final status.

A real-world example

A mid-sized commercial printing company runs Sage 100 on-premises for AR, AP, and GL. Their banking team manually logs into their business bank account three times a week to download transactions, then re-enters them into Sage 100 for reconciliation. Separately, the AP team creates checks in Sage 100 but must export approved batches and transfer the money through online banking, a two-system workflow that creates data drift and manual error. With Sage 100 and Plaid connected, bank transactions appear in Sage 100 automatically each night, reconciled to GL accounts via mapping rules. Approved AP payments in Sage 100 trigger ACH transfers directly through Plaid, with transfer status flowing back into Sage 100. The finance team spends two hours per week less on transaction entry and payment processing.

What you can do

• •Pull bank transactions from Plaid into Sage 100 for automatic GL reconciliation, mapped to accounts via customer-configured rules.
• •Initiate ACH and wire transfers from Sage 100 approved payments, routed through Plaid with full audit trail and status tracking.
• •Verify Plaid webhook signatures using JWT ES256 decoding and timestamp validation to prevent spoofed transaction notifications.
• •Store Plaid access tokens and Sage 100 credentials encrypted, with idempotency keys on transfers to prevent duplicate payments on retry.
• •Poll Sage 100 and Plaid on a schedule independent of each system, with exponential backoff on rate-limit errors and full replay on downstream failure.

Questions

What data moves between Sage 100 and Plaid?

Bank transactions and balances move from Plaid into Sage 100 for GL reconciliation. Approved vendor payments in Sage 100 initiate ACH transfers through Plaid. Plaid transfer status flows back into Sage 100 to update payment state. Customer invoices and AP invoices do not flow to Plaid since Plaid is a bank data network, not an ERP.

Does ml-connector require changes to Sage 100 or Plaid setup?

Sage 100 requires a Windows service account with access to eBusiness Web Services or the COM layer, and the Sage user must be individually enabled for Web Services in Sage 100 admin (off by default). Plaid requires a Plaid Dashboard account with API keys and the transactions and transfer products enabled. ml-connector stores and rotates the Plaid access token obtained via Plaid Link, so the user logs in once to authorize.

How does ml-connector handle bank reconciliation if a transaction is missing or late?

Plaid transactions are fetched with up to 24 months of history and include a transaction_id that does not change. If a transaction arrives late or is corrected by the bank, ml-connector checks for duplicate transaction_id before inserting a journal entry into Sage 100. If an entry already exists, it is skipped. Corrections from Plaid are handled as separate transactions, allowing the finance team to review and match them manually if needed.