Campaign Agents
One experiment each — an audience slice, a channel and a content angle — run on demand or on the daily plan, and measured per agent.
A campaign agent is one experiment: one audience slice × one channel × one content angle, aimed at a product and a goal. It is the thing you create, run, compare and retire. Setup seeds the first few, paused; running one produces a batch of drafts you review together, so you can tell which slice is working from a cost and a click rate you can read per agent.
Run one
curl --fail-with-body "https://moabcrm.com/api/v1/campaign-agents/$AGENT_ID/runs" \
-H "Authorization: Bearer $MOAB_API_KEY" \
-H "Project-ID: $PROJECT_ID" \
-H 'Content-Type: application/json' \
-H 'Idempotency-Key: acme-campaign-agents.run-1' \
-d '{"model":"meta/muse-spark-1.3"}'The answer is the task acknowledgement — poll GET /tasks/{id} for the
receipt. A run researches leads for the agent's slice and writes drafts, all
in status: "draft"; it never sends. Pass model to pin the model for this
run; omit it for the governed default.
Approve it for ongoing work
curl --fail-with-body "https://moabcrm.com/api/v1/campaign-agents/$AGENT_ID/approve" \
-H "Authorization: Bearer $MOAB_API_KEY" \
-H "Project-ID: $PROJECT_ID" \
-H 'Content-Type: application/json' \
-H 'Idempotency-Key: acme-campaign-agents.approve-1' \
-d '{}'Approving an agent moves it to status: "live" — the same press as the
product's Live switch — so the daily plan may run it and its approved drafts
may send. Its product and goal go live with it. A paused agent (setup's
default) only runs when you run it.
Read what it earned
curl --fail-with-body "https://moabcrm.com/api/v1/campaign-agents/$AGENT_ID/performance?metric=links_clicked&bucket=day" \
-H "Authorization: Bearer $MOAB_API_KEY" \
-H "Project-ID: $PROJECT_ID"metric is any of the dashboard's measures (links_clicked by default,
cost_usd, replies, opens); bucket is day or total. The same read
across the whole project is Analytics.
Create one by hand
Create a campaign-agent — POST /api/v1/campaign-agents
curl -X POST "https://moabcrm.com/api/v1/campaign-agents" \
-H "Authorization: Bearer $DASH_API_KEY" \
-H "Project-ID: pr_zest" \
-H "Content-Type: application/json" \
-d '{"content_id":"$CONTENT_ID","product_id":"$PRODUCT_ID","goal_id":"$GOAL_ID","employee_id":"$EMPLOYEE_ID","provider":"instantly","icp_axes":{"industry":"cybersecurity","role":"CTO","headcount":"51-200"},"status":"paused"}'Add an agent for CTOs at cybersecurity companies with 51-200 peopledata_set {
"collection": "agents",
"object_id": "draft-001",
"data": {
"content_id": "$CONTENT_ID",
"product_id": "$PRODUCT_ID",
"goal_id": "$GOAL_ID",
"employee_id": "$EMPLOYEE_ID",
"provider": "instantly",
"icp_axes": {
"industry": "cybersecurity",
"role": "CTO",
"headcount": "51-200"
},
"status": "paused"
}
}{
"object": "campaign-agent",
"id": "obj_123",
"name": "Cybersec CTO/VPE (51-200) × Resend",
"status": "live",
"kind": "outbound",
"reactivation_after_days": 1,
"note": "25 people found; wrote 20 emails from the golden piece.",
"icp_axes": {
"industry": "cybersecurity",
"role": "CTO",
"headcount": "51-200"
},
"program": "program",
"product_id": "$PRODUCT_ID",
"goal_id": "$GOAL_ID",
"content_id": "$CONTENT_ID",
"list_ids": [],
"parent_id": "agent-ionwarp-cybersec-vpeng",
"axis": "role",
"born_by": "planner",
"tags": [],
"provider": "instantly",
"channel": "email",
"employee_id": "$EMPLOYEE_ID",
"outreach": {
"messaging": "messaging",
"follow_up_bodies": [],
"linkedin_fallback_message": "linkedin fallback message"
},
"created_at": "2026-07-25T12:00:00.000Z",
"updated_at": "2026-07-25T12:00:00.000Z"
}The resources
campaign-agents— the experiments themselves.campaign-agent-runs— read-only. One batch of work.reports— the day's report record: what ran, what it produced, and what needs you.
Worth knowing
- A Moab campaign is a campaign agent — there is no separate campaign
object, and a batch of drafts is a run: drafts group by the
run_idof the run that produced them, so you review a run rather than a stream. - Autopilot is already on. One scheduled planner per project wakes once a
day and splits the day's budget across the live agents;
POST /campaign-agents/runsruns that planner on demand. Its caps (moab.planner.daily_review_draft_cap,daily_send_cap,daily_new_agent_cap,daily_new_content_cap) are per-project configuration. - An agent whose channel no connected provider can serve is refused by name at write time, not quietly ignored later. That refusal and the approval brake that decides whether an autonomous run may start are on Control plane.
Next: Outbox — what a run leaves you to approve.