ElevenLabs launches Campaign Manager for outbound AI agents
ElevenLabs launched ElevenAgents Campaign Manager on November 9, 2026, giving businesses a way to coordinate outbound AI conversations across repeated attempts. Phone calls, WhatsApp voice and WhatsApp messaging are available now. Email and SMS are still planned.
The practical change is continuity. A missed call, a requested callback and a completed task can become stages in one managed campaign. For teams evaluating the launch, the important questions concern how contact stops, how exceptions are handled and which records persist.
What Campaign Manager adds
According to the announcement, teams can reuse their existing agents and phone numbers, enroll recipients through CSV or API, and track progress from enrollment to a configured outcome. A contact history explains the next action, while uncertain cases enter a queue for human attention. Results can be exported to CSV.
Existing customers can find Campaigns in the Agents sidebar. ElevenLabs directs new customers to sales. The launch post does not establish a new price or detailed plan eligibility. Its customer results are testimonials, not independent measurements of campaign effectiveness.
Choosing a voice system remains a separate decision. Our Eleven Realtime guide covers the model and integration requirements behind that choice. A campaign scheduler does not by itself establish whether an agent can correctly finish your task.
The controls have boundaries
The campaign documentation says campaigns replace batch calling. Phone campaigns support retries, callbacks and redirects. WhatsApp text campaigns send approved templates without retries.
Several operational details deserve a check before rollout:
A campaign keeps the policy set when it was created. Changing the workspace default does not change an existing campaign. Recipient requests for an exact callback time can override its normal calling windows.
Do not call state belongs to each campaign. Teams must deduplicate recipients across campaigns rather than assume a stop request automatically applies everywhere.
Campaigns are unavailable for workspaces or agents using Zero Retention Mode. Follow-up decisions require durable call state and conversation analysis. The documentation does not establish a requirement to keep complete transcripts permanently.
These boundaries matter in an ordinary service workflow. If a customer asks to stop receiving calls, a second campaign should not quietly contact them again. Before enrollment, decide where suppression requests are recorded and how that record reaches every campaign.
WhatsApp needs its own preparation
ElevenLabs’ WhatsApp outbound guide requires approved templates when a business starts a conversation. The agent waits for a reply before continuing. An accepted API request does not guarantee delivery by Meta.
Outbound WhatsApp calls also require recipient permission. Meta’s template pricing and messaging limits continue to apply. Channel availability therefore should not be read as permission to contact an arbitrary list.
Prepare the consent record, template approval and recipient list before running a trial. Keep delivery failures separate from conversations that reached a person but failed to finish the task. Those are different problems and need different fixes.
Start with a small, measurable trial
For a first evaluation, use a short, authorized list and a clear definition of success. Include unanswered calls, ambiguous callback requests, duplicate contacts and explicit stop requests. Check the human review queue as carefully as the completion count.
Our recommendation is to measure correct outcomes, unwanted repeat contact, human handoffs and total cost together. We have not tested Campaign Manager. Its value will depend on how reliably the surrounding workflow honors the recipient’s request, not simply how many conversations it starts.
