AI reply drafts and the Setter

When a reply deserves an answer, the agent can write it — but who presses send is a decision you make explicitly, and can scope campaign by campaign.

Reply drafts: you approve every send

With drafts enabled, the agent prepares replies in your tone, grounded in the conversation. Each draft waits in the Pending drafts view with three actions: approve, edit, or reject (with an optional reason the agent learns from).

Two modes:

• On demand — you press “draft a reply” in a conversation and get one, when you want it.
• Proactive — the agent prepares drafts as replies arrive, within a daily cap you set, in the language you choose.

Drafts are safe by construction: a draft expires on its own if the prospect replies again in the meantime — you never approve an answer to an outdated conversation.

The Setter: delegated replies, scoped by you

The Setter is the optional step further (a paid add-on): on the campaigns you designate, the agent sends its replies itself and only escalates the conversations that are ready to close — a proposed meeting, a hot discussion that needs a human.

The scope rules are strict by design:

• Per campaign, opt-in — never a workspace-wide switch, never a default.
• You can step back to approval mode at any moment.
• Every sent reply is visible in the conversation and in the journal, attributed to the agent.

Which mode for whom

Start with drafts and approvals — it costs you seconds per conversation and builds the trust the Setter needs. Delegate to the Setter the campaigns where conversations are homogeneous and the stakes of a single reply are low; keep approval on the audiences where every word matters.