How does ml-connector authenticate against Sage 100 if it is on-premises and has no OAuth?

ml-connector uses the local Sage 100 agent endpoint that the customer runs on their Windows server. It passes Windows service account credentials and an agent-layer API key (or mTLS certificate) to the agent, which handles authentication to the BOI COM layer. The customer must enable Web Services in Sage 100 admin and provide the agent endpoint URL and authentication credentials to ml-connector.

Why does ml-connector poll Sage 100 instead of waiting for a push?

Sage 100 has no webhooks or event stream capability, so polling is the only option. ml-connector polls on a schedule you define (recommended: customers hourly, invoices every 15 minutes) using DateLastUpdated and DateCreated fields to fetch only records that have changed since the last poll.

What happens if a Sage 100 customer exists in both systems - how does ml-connector avoid duplicates?

ml-connector checks for existence in HubSpot before creating a new contact or company, using the customer number or external ID to find matches. If a record already exists, ml-connector updates it with the latest data from Sage 100 rather than creating a duplicate.

Sage 100HubSpot

Sage 100 and HubSpot integration

Sage 100 holds your AR customers, invoices, and vendor data. HubSpot holds your sales pipeline and customer relationships. Connecting them keeps your sales team aligned with your revenue: new customers created in Sage 100 appear as prospects in HubSpot, outstanding invoices are visible to the sales and support teams, and vendor information can be synced back into company records. ml-connector handles the polling model and on-premises authentication required by Sage 100's architecture.

How Sage 100 works

Sage 100 is an on-premises ERP system covering accounts receivable, accounts payable, general ledger, inventory, purchasing, and sales. Data is accessed through SOAP web services (limited to Sales Orders and Customers AR only) or through a local Windows agent that wraps the BOI COM layer, which provides full access to customers, invoices, vendors, purchase orders, and GL accounts. Authentication uses username and password per SOAP call, or Windows service account credentials plus agent-layer authentication. Sage 100 has no webhooks or event stream, so polling is required. Polling is done by checking DateLastUpdated or DateCreated fields, with recommended intervals of 15 minutes for AP invoices and POs, hourly for vendors, and daily for GL accounts.

How HubSpot works

HubSpot is a cloud-based CRM and sales platform that exposes contacts, companies, deals, invoices, line items, products, payments, and orders through a REST API at https://api.hubapi.com. Authentication uses Bearer tokens, either a Private App Access Token (pat-na1- format) for server-to-server integrations, or OAuth 2.0 for public apps. HubSpot supports both webhooks with signature verification and polling via list/search endpoints. Webhooks require a Public App registration separate from Private App credentials. The search API is limited to 10,000 records; larger datasets require the Lists API or CRM Export. Batch operations are capped at 100 records per request, and rate limits vary by tier: Free and Starter at 100 per 10 seconds, Professional and Enterprise at 190 per 10 seconds.

What moves between them

The main flow is from Sage 100 into HubSpot. Sage 100 AR customers map to HubSpot contacts and companies, with annual revenue and industry sourced from Sage 100 GL and customer records. Sage 100 invoices map to HubSpot invoices, associated with the matching customer contact and linked to deals where applicable. The sync runs on a schedule: customers and vendors every hour, invoices every 15 minutes. Payment and AR aging data flow from Sage 100 into HubSpot company records so the sales team can see which customers have outstanding balances and payment terms. This is a pull-only flow; Sage 100 is the system of record for financial data.

How ml-connector handles it

ml-connector uses the local Sage 100 agent endpoint provided by the customer, along with Windows service account credentials and an agent-layer API key for BOI access. It polls Sage 100 on the schedule you define, reading customers from the AR_Customer entity, invoices from AR_Invoice, and vendors from AP_Vendor. Each record is checked for existence in HubSpot before create, preventing duplicates. Sage 100 records are mapped to HubSpot field names (e.g., CustomerNumber to externalId, CustomerName to firstname/lastname). Because Sage 100 has no idempotency keys, ml-connector uses the external ID plus timestamp to detect re-enqueued polls. The local agent connection requires the customer's IIS certificate and hostname binding to be correct; certificate validation failures are surfaced in the audit log with the WSDL endpoint for debugging. Company codes (the 3-character identifier required by every Sage 100 call) are specified per Sage 100 customer instance in ml-connector configuration. Rate limiting against HubSpot is handled by queuing and backoff, and the full audit trail lets you replay any record if a sync fails downstream.

A real-world example

A mid-market B2B software company runs Sage 100 for accounting and AR, and HubSpot for sales and customer success. Before the integration, the sales and support teams exported AR aging reports from Sage 100 manually and pasted them into a shared spreadsheet to track which customers had outstanding balances. New customers created in Sage 100 were not automatically visible to HubSpot, so the sales team often lost visibility into which accounts in the CRM were actually customers. With Sage 100 and HubSpot connected, AR customers and their invoices sync hourly, so the sales team can see outstanding balances and payment terms directly in HubSpot company records. New customers are created as HubSpot companies with AR data pre-populated, and invoices appear as linked records so support can proactively follow up on aging AR.

What you can do

Sync Sage 100 AR customers into HubSpot as contacts and companies with revenue and industry data from the ERP.
Pull Sage 100 invoices into HubSpot on a 15-minute schedule and link them to customer company records.
Map Sage 100 vendors to HubSpot companies so procurement and supplier data is visible in the CRM.
Authenticate against Sage 100's local Windows agent using service account credentials and agent-layer API keys.
Poll on a configurable schedule with full audit trail, deduplication, and replay capability for failed syncs.

Questions

How does ml-connector authenticate against Sage 100 if it is on-premises and has no OAuth?: ml-connector uses the local Sage 100 agent endpoint that the customer runs on their Windows server. It passes Windows service account credentials and an agent-layer API key (or mTLS certificate) to the agent, which handles authentication to the BOI COM layer. The customer must enable Web Services in Sage 100 admin and provide the agent endpoint URL and authentication credentials to ml-connector.
Why does ml-connector poll Sage 100 instead of waiting for a push?: Sage 100 has no webhooks or event stream capability, so polling is the only option. ml-connector polls on a schedule you define (recommended: customers hourly, invoices every 15 minutes) using DateLastUpdated and DateCreated fields to fetch only records that have changed since the last poll.
What happens if a Sage 100 customer exists in both systems - how does ml-connector avoid duplicates?: ml-connector checks for existence in HubSpot before creating a new contact or company, using the customer number or external ID to find matches. If a record already exists, ml-connector updates it with the latest data from Sage 100 rather than creating a duplicate.

Connect Sage 100 and HubSpot

Free to use. Add your credentials, ping your real systems, and see if we fit.

Get started