An agent that runs the campaign operations loop end to end, then recommends a content agent to draft copy from the same campaign brief. Built on Claude and Python across HubSpot and Salesforce.
Campaign operations is a chain of manual, easy-to-fumble steps: build the assets, check the data underneath them, sequence the launch, and reconcile what actually shipped against what was briefed. The work is repetitive, it is slow, and every handoff is a place for a good campaign to stall.
MOPSbot reads the campaign brief, validates the data before anything is built, sequences the work in the right order, and reports back what changed. Its standout move is recommending a content agent to draft the campaign from that same live context, so content starts from real data instead of a blank page.
Claude sits at the center as the reasoning layer; Python carries the work across the HubSpot and Salesforce APIs. The assistant is built the way I build everything: ship a working slice, watch it run against real campaigns, then optimize and iterate on the parts that earn it.
The full write-up is in progress. The summary above is live now.