AI Reply Setter is planned to help teams continue LinkedIn conversations after a prospect responds. Copilot prepares replies for approval; Autopilot can continue eligible conversations using a configured objective and sender style.
Production access is planned in stages. This guide is a preparation workflow, not a promise that the controls are enabled in every workspace. Confirm your access and the demonstrated behavior with the OutFlo team before relying on automation.
Choose a mode and a responsible person
- Mode
- Off
- Intended workflow
- No setter automation for the campaign
- Team decision
- Who handles replies manually
- Mode
- Copilot
- Intended workflow
- AI drafts wait for approval
- Team decision
- Who reviews, edits and approves them
- Mode
- Autopilot
- Intended workflow
- Eligible conversations can continue automatically
- Team decision
- Which cases are suitable and who monitors them
| Mode | Intended workflow | Team decision |
|---|---|---|
| Off | No setter automation for the campaign | Who handles replies manually |
| Copilot | AI drafts wait for approval | Who reviews, edits and approves them |
| Autopilot | Eligible conversations can continue automatically | Which cases are suitable and who monitors them |
Copilot is useful when replies require careful judgment or your team is still checking generated quality. Autopilot needs a narrower, well-understood objective and a clear handoff process. These are operational recommendations, not a claim that either mode guarantees meetings.
Assign an owner who can review conversations and take over. Do not enable a mode just because a campaign has many replies.
Prepare sender style and a specific objective
Collect writing examples that represent the sender's tone, sentence length and preferred next steps. Provide only information that belongs in that workflow; do not depend on automatic style learning.
Define an objective such as “Confirm whether the prospect owns outbound operations, answer questions using approved product information, and offer the calendar link when they ask to speak.” That gives the responder a clearer task than “book more meetings.”
Prepare accurate answers about your offer and decide which questions require a teammate. A calendar link can be shared where configured, but native calendar booking is outside the planned release. Sharing a link does not mean an appointment was booked.
Understand which conversations can enroll
The intended live flow depends on unambiguous attribution to an enabled campaign, sender and prospect after activation. A historical conversation does not automatically become an enrolled conversation; it needs explicit enrollment.
Review sample threads from your campaign. Check which campaign owns each one, whether the sender is correct and whether a person is already managing it. Ambiguous attribution is a reason to inspect the context, not to assume the AI will choose correctly.
Plan no-response follow-ups
The documented workflow allows up to three configured follow-ups, each with its own delay, purpose and supported filters. Choose text or voice according to the enabled capabilities.
For example, you might plan one reminder after an unanswered question and a final close-out message later. More available slots do not make more follow-ups appropriate. Decide what stops the flow and when your team should take over.
Voice follow-ups require a ready sender voice and the relevant access and credits. Unavailable voice should not be treated as a promise of an automatic text replacement. Incoming audio transcription is outside this release, so plan human handling for conversations that depend on understanding an audio reply.
Agree on human takeover and delivery review
Human takeover remains in effect until an explicit resume. Tell teammates how to identify an AI-managed conversation, take control and decide whether it should return to AI later.
Prospect opt-outs block setter resume. Operating hours, sender quotas and configured monthly credit budgets also constrain the workflow.
If delivery is uncertain, inspect and reconcile the outcome before retrying. A message that may already have reached the recipient should not be resent simply because its status is unclear. Follow the enabled reconciliation controls and involve support if you cannot determine the result.
Review quality before expanding use
For an initial enabled cohort, review actual drafts and conversation outcomes. Check factual accuracy, sender voice, whether questions are answered, inappropriate follow-ups and when human takeover occurs. Inspect budget consumption as well as message volume.
The planned release does not include arbitrary imported skills, native meeting booking or incoming audio transcription. Keep the workflow within the capabilities shown for your workspace.
For current human-reviewed inbox assistance, see OutFlo Agent. For staged-release access, explore AI Reply Setter.