Sage 100 and Asana integration

What moves between them

The main flow runs from Asana into Sage 100. When a purchasing task reaches an approved state in Asana, ml-connector reads its custom fields for the vendor, cost code, and line detail and creates a purchase order in Sage 100 through the local agent, against a valid vendor and segmented GL account. The reverse flow writes the new Sage 100 purchase order number, vendor name, and posting status back into the Asana task's custom fields so the requester sees progress. The Asana user roster is read so requesters and approvers can be matched to Sage 100 records. Asana task webhooks trigger the work in near real time, and a scheduled poll of Sage 100 picks up status changes since Sage 100 cannot push.

How ml-connector handles it

ml-connector stores both credential sets encrypted. For Asana it sends the bearer token on every call and requests opt_fields so custom fields come back populated rather than as sparse objects. For Sage 100 it calls the customer's local Windows agent over HTTPS, which wraps the Business Object Interface, and it passes the company code and APDivisionNo on every purchase order write. Because Asana custom field GIDs are workspace-scoped and can be renamed, fields are mapped by GID, not by name. Asana provides no idempotency key, so the connector stores the Asana task GID and the Sage 100 purchase order key and skips re-creating one that already exists. The Asana webhook is verified by recomputing the HMAC-SHA256 signature, and because Asana silently deletes a webhook after 24 hours of failed delivery the connector periodically checks that it still exists and re-registers it. Sage 100 record-locking errors are retried with backoff, and every record carries a full audit trail and can be replayed.

A real-world example

A regional construction firm of about 200 staff runs Sage 100 on a server in its head office for purchasing and accounts payable, and runs project work and field requests in Asana. Before the integration, a project manager filled out an Asana task to request materials, an approver checked a box, and then someone in accounting re-typed the whole thing into Sage 100 as a purchase order, often days later and sometimes against the wrong cost code. With Sage 100 and Asana connected, an approved Asana request becomes a Sage 100 purchase order automatically, mapped to the right vendor and job cost account, and the PO number flows back to the task. Approvers see the order is placed, and accounting stops re-keying requests by hand.

What you can do

• •Turn an approved Asana purchasing task into a Sage 100 purchase order through the on-prem local agent.
• •Write the Sage 100 purchase order number, vendor, and posting status back to the Asana task's custom fields.
• •Map Asana custom fields to Sage 100 vendors, cost codes, and segmented GL accounts by GID.
• •Bridge Asana bearer-token auth to Sage 100 username, company code, and local agent API key.
• •Trigger on Asana task webhooks, poll Sage 100 for status, and keep a full audit trail with replay.

Questions

Can Asana create purchase orders or invoices in Sage 100 directly?

Asana has no native purchase order, invoice, or GL account objects, so the request lives as a task with custom fields. ml-connector reads an approved task's custom fields and creates the actual purchase order in Sage 100 through the local agent, which reaches the Business Object Interface. The financial record is created and validated in Sage 100, not in Asana.

Why does Sage 100 need a local agent for this integration?

Sage 100 is on-premises and has no cloud API, and full access to vendors and purchase orders is through the Business Object Interface, a COM layer that only runs on the Sage 100 server. A lightweight Windows agent on that server exposes those operations over HTTPS so ml-connector can reach them. Without the agent only sales orders and customers are reachable through the SOAP service.

How are updates kept current when Sage 100 cannot push events?

Asana sends a signed webhook when a task changes, so requests move into Sage 100 in near real time. Sage 100 has no webhooks, so ml-connector polls it on a schedule using the DateLastUpdated fields to pick up posting status. Because Asana deletes a webhook after 24 hours of failed delivery, the connector checks the webhook still exists and re-registers it when needed.