Pause the AI Agent for a Contact
Temporarily stop automated replies for one contact when a person should control the conversation—without disabling the AI Agent for everyone else.
Pause the contact, then give the conversation an owner
Open the conversation, review the current AI status and use the available pause control. Assign the conversation or create a task, add only the context the next person needs and keep the AI paused until the human issue is complete. Before resuming, read everything sent during the pause and confirm the Agent can continue safely.
Use a contact-level pause when a person should take control
Pause one contact when the current conversation needs human judgment, approved handling or a manual exchange that the AI Agent should not interrupt.
The customer asks for a person
Respect the request and route the conversation to an authorized team member.
Pause before the next automated replyComplaint or sensitive issue
A complaint, policy exception or sensitive situation needs careful human review.
Preserve the relevant contextTeam member, partner or test record
The contact should not receive normal automated customer responses.
Use the correct long-term control if neededHigh-value or unusual situation
A specific person is responsible for managing the conversation directly.
Document ownership and next actionThe Agent lacks approved information
The request falls outside its current knowledge or response boundaries.
Fix and test before resumingYour team is already replying
The AI Agent should remain quiet while a person conducts the active conversation.
Avoid overlapping responsesUnderstand what changes—and what continues
While paused, the AI Agent remains quiet for that contact. Incoming messages can still be stored in the inbox, team notifications may continue and an authorized person can review and reply according to the channel’s rules.
Do not confuse pausing with archiving or blocking
Pausing the AI is different from archiving, blocking or do-not-contact status. Choose the control that matches the customer’s request, the state of the work and your organization’s communication policy.
Stops automated replies for the selected contact while a person handles the conversation.
Temporary and contact-specific; incoming messages can still be available.
Closes or removes current inbox work from the active view.
Does not necessarily change automation or future communication permissions.
Stops defined communication with the contact or channel identity.
Use only when blocking matches the requested or policy-defined outcome.
Records a communication restriction or consent decision.
Do not replace this requirement with a temporary AI pause.
Review, pause and preserve ownership
The exact placement of the pause control can vary by workspace configuration. Work from the contact’s active conversation and verify the resulting AI status before leaving the record.
- Open the correct conversation.Confirm the contact identity, channel and active thread before changing the status.
- Review the current context.Read the latest messages and determine why a person should take control.
- Use the available pause control.Confirm that the AI status changes for this contact.
- Assign the work.Name the responsible person or team and create a task when a timed next step is required.
- Add concise context.Record the reason, current commitment and next action without copying unnecessary private detail.
- Verify the handoff.Make sure the owner can access the conversation and understands when to respond.
Reply manually within the channel’s rules
An authorized person can continue the conversation while the AI Agent is paused. Before replying, confirm the correct contact, channel, sender identity, current service window when applicable and any approved template or consent requirement.
The message is going to the intended contact from the approved sender.
Replying from the wrong conversation, number or account.
The owner has read the current thread and any relevant assignment notes.
Contradicting an earlier promise or asking the customer to repeat information.
The reply is allowed by the customer’s choices and the channel’s current rules.
Treating an AI pause as permission to send an otherwise restricted message.
The response resolves the issue or states the next step, owner and timing.
A manual exchange that ends without a clear outcome.
Confirm the filter and count before pausing several contacts
When your account supports pausing selected contacts together, treat it as a controlled operational change. Verify the exact selection and downstream coverage before applying it.
Build the intended group
Confirm selection size
Review contacts
Apply one change
Spot-check results
Confirm that enough authorized people can monitor and respond to the paused group.
Decide how each paused conversation receives a responsible owner and next action.
Check included and excluded records rather than relying only on the filter name.
Define whether contacts will resume individually or after a verified shared condition.
Read the paused period before reactivating the AI Agent
Resume intentionally. The Agent may evaluate the conversation and determine whether a reply or follow-up is appropriate after it becomes active.
- Read every new message.Review the complete exchange that occurred while the AI Agent was paused.
- Confirm the human issue is complete.Do not resume while a complaint, exception, negotiation or named-owner commitment remains active.
- Check promises and exceptions.Make sure nothing a person said conflicts with the Agent’s current instructions or next-step logic.
- Refresh the context.Update the appropriate field, note, tag, task or approved knowledge source when the conversation changed the record.
- Test any exposed gap.If the pause revealed missing knowledge or unsafe behavior, correct and test that scenario before resuming.
- Use the resume control.Confirm the contact’s AI status and continue monitoring the next relevant interaction.
Fix the reason for the pause when it reveals a system gap
A pause can surface missing knowledge, unclear boundaries, weak escalation behavior or a situation the Agent should recognize earlier. Preserve a sanitized example and improve the relevant source before returning that scenario to normal traffic.
What made human control necessary?
Record the missing information, unsafe response risk, unclear intent or escalation trigger using a synthetic or sanitized example.
Update and retest the exact scenario
Change the approved knowledge, instruction or workflow, then test direct and edge-case versions before resuming.
Add an approved source
Provide accurate, current guidance and define what the Agent should do when the answer is still unavailable.
Tighten the escalation rule
Make the human handoff condition explicit and test that the Agent stops rather than improvising.
Align instructions and context
Update the relevant record or workflow so the Agent does not contradict a valid human commitment.
Review the broader workflow
If the same pause reason appears repeatedly, correct the shared source instead of treating every contact as an isolated exception.
Preserve only the context needed for the handoff
Paused conversations often involve unusual or sensitive situations. Limit access to authorized people and record only what the next person or approved workflow genuinely needs.
Minimum context
State the pause reason, current commitment and next action without copying unnecessary private detail.
Authorized owner
Assign only people who should review and respond to the conversation.
Protected evidence
Keep customer information in the approved conversation or record rather than in uncontrolled working copies.
Consent controls
Preserve opt-outs, restrictions and channel-specific permissions independently of the AI pause.
Resolve common pause and resume problems
Verify the contact and current status
Confirm the exact conversation, channel and contact record, then check whether another approved automation or sender produced the message.
Inspect assignment and task ownership
Confirm a named person can access the conversation, received the notification and has a clear due time.
Check access and conversation state
Verify the user’s permissions, the current AI status and whether another contact or channel control takes precedence.
Review the paused-period context
Check human messages, promises, notes, fields and Agent knowledge before resuming again.
Stop and preserve the selection evidence
Record the filter, count and sample contacts, correct the affected statuses and verify ownership before another change.
Correct the shared knowledge or boundary
Use sanitized examples to update and test the Agent instead of repeatedly relying on manual intervention.
Pausing the AI Agent questions
Does pausing one contact stop the AI Agent for everyone?
No. A contact-level pause keeps the Agent quiet for that contact while other eligible conversations continue under their existing settings.
Will incoming messages disappear while the AI is paused?
Incoming messages can still be stored in the inbox, and team notifications may continue. Confirm the account’s current channel and notification settings.
Can a person reply while the AI Agent is paused?
Yes. An authorized person can review and reply according to the channel’s rules, the customer’s choices and your organization’s policy.
Is pausing the same as archiving or blocking?
No. Pausing stops the AI Agent for the contact. Archiving closes current inbox work, while blocking or do-not-contact controls stop defined communications.
Can I pause several contacts at once?
When supported by the account, selected contacts can be paused together. Confirm the filter and count, review a representative sample and spot-check the result.
When should I resume the AI Agent?
Resume after the human issue is complete, paused-period messages have been reviewed and no promise, exception, knowledge gap or consent restriction conflicts with the Agent’s behavior.
Send the conversation status and expected behavior
Include the ConvoRev account, contact and channel, approximate time, current AI status, expected result and a sanitized example. Remove credentials and unnecessary private customer information.