Start the conversation. Keep it useful. Bring in the agent when it matters.
Autopilot finds the contacts your team isn't touching, writes each of them something real in the agent's own voice, continues verified replies, and brings in the agent for a qualified next step or a human request. Drafts-first by default — you decide when it earns the send.
Included with Ace Trove · Drafts-first by default · Check activation and effective permissions in admin
What is Ace Autopilot? Ace Autopilot is Follow Up Ace's autonomous AI outreach system for Follow Up Boss, using authorized CRM context and current enrollment and policy checks. It runs one system of configurable workflows for new leads, re-engagement, and nurture, sharing account controls, draft review, and conversation handoff. It selects the new and resurfacing leads no one on the team is working, composes a personalized touch grounded in something real about that specific contact, runs every message through a Fair-Housing compliance scan and enrollment checks, and starts in drafts-first mode — each touch lands in the assigned agent's one-tap queue until an admin graduates the playbook to send on its own. When current message content and origin can be verified, Ace continues the conversation. A qualified next step or an explicit human request brings in the assigned agent with the available context; unavailable content and uncertain outcomes require review. Email sends from each agent's own Gmail; SMS is available through a team's own Lead Engage subscription. Included with Ace Trove.
Select → write → check → converse → hand off.
Every team has leads going cold in plain sight: the inquiry from Tuesday nobody opened, the buyer from last spring who just started looking again. Autopilot runs a disciplined loop over exactly those contacts — on the platform Ace already runs for you: the live webhook feed from Follow Up Boss, the lifecycle engine, the predictive models, and seller intent decide who and when; one content engine writes what.
- Select. Ace already watches your whole database — lifecycle state, resurfacing evidence, model scores, seller intent, engagement. Autopilot picks the contacts in a playbook's lane that nobody is touching, and refuses everyone it shouldn't write (more on that below — the refusals are the feature).
- Write. Each touch is composed for that one person, from things that are actually true: how they reached you, available first-party capture answers, recorded referrals and notes the agent recorded — in the agent's own voice, with real, dated market context when it helps.
- Check. Every message passes the same Fair-Housing compliance scan that guards the rest of Ace, enrollment and opt-out checks, the account's configured sending-hours window, daily caps, and carries an AI-assistance disclosure. The enrollment tag does not override opt-outs or sending limits.
- Deliver. Drafts-first by default — into the agent's one-tap queue. Graduated playbooks send as the assigned agent, from their own connected Gmail. On the SMS bridge, your own texting subscription delivers.
- Converse and hand off. Verified replies can continue as a grounded conversation. A qualified next step or an explicit request for a human brings in the assigned agent with a summary and recorded context. Opt-outs, human takeover and an under-contract state stop contact automation. An alert being recorded does not prove the agent received it.
One Autopilot workspace: Activity shows outcomes, Workflows configures new leads, re-engagement, and nurture, and Controls sets targeting, review, and actions. Default plans are finite: new leads on days 0, 3, and 7; re-engagement on days 0 and 7; nurture on days 0, 7, and 21. Sending windows, review, and eligibility can delay a touch.
It drafts before it sends. You decide when it graduates.
Most "AI follow-up" products ask you to trust them on day one. Autopilot is built the other way around:
- Drafts-first is the default. The entire loop runs — selection, composition, compliance, enrollment — but every touch lands as a draft in the assigned agent's existing one-tap queue. You watch what Autopilot would have sent before it ever sends.
- Graduation is per playbook, and reversible. An admin turns on one motion at a time — new-lead first touch, resurfacing re-engagement — and can send any of them back to drafts-first with one switch. Nothing sends on its own until you say so.
- Budgets are hard ceilings. Account-wide and per-playbook daily caps are enforced atomically — an over-busy Autopilot is a configuration you chose, never a runaway.
Choose the actions and knowledge. Admin can request Off, Draft or Execute for replies, allowed stage transitions, tasks, notes, selected sequences and partner call requests. The effective mode also depends on the playbook's review ceiling, Ace Trove and current capability checks. Appointment preferences offer Off or Draft only.
Knowledge can be off, all eligible account knowledge, or selected entries. Global AI exclusions and approval rules still apply, and knowledge never grants permission to act. Revoked or changed knowledge is checked again before a new action or send.
What makes it trustworthy is everything it refuses to do.
Autopilot's selection is conservative by construction, and every "no" is recorded with its reason so an admin can see exactly why a contact wasn't touched.
-
Start with a tag. Choose Only contacts I tag, then add
ace-autopilotin Follow Up Boss. That is explicit enrollment, with no additional contact consent artifact. Theace-autopilot-excludeandai_deadtags stop every modern Autopilot workflow, including nurture. Whole-book mode retains its relationship-evidence requirement. - Nothing real to say, no message. If Ace holds nothing specific about that person to ground a message in, it refuses — because a message with nothing real to say is a generic drip, and generic drips train your leads to ignore your agents' real mail too.
- Your agents' momentum is never interrupted. A contact the assigned agent touched in the last 7 days is off-limits — no score or signal overrides that. Autopilot works the leads nobody is working, full stop.
- Opt-outs keep their scope. Full opt-outs stop all outreach. A channel-only texting stop closes SMS; email still needs its own eligibility and policy checks. These checks run before composition and again at send time.
- It never talks over a live conversation. Under contract, mid-thread with a human, or already being worked by another assistant — Autopilot stands down.
The compliance layer underneath — the Fair-Housing scan, the AI disclosure, quiet hours — is shared with the rest of Ace. See how Ace handles compliance.
Email from your agents' own inboxes. SMS on your own subscription.
- Email — native. A graduated touch sends as the contact's assigned agent, from that agent's own connected Gmail — the address the lead already knows, with replies landing in the agent's real inbox. No connected sender, no send: Autopilot refuses rather than mailing from a platform address.
- SMS — bring your own texting platform. Ace itself never sends a text. Teams with their own Lead Engage subscription can turn on the Autopilot SMS bridge: Ace composes each message and lays the schedule into Follow Up Boss, and their platform delivers from your number. First texts identify the agent and carry "Reply STOP to opt out"; verified replies can continue the conversation. Opt-outs and human takeover stop local automation and create tracked provider cleanup. Ace uses authorized account context to draft, then sends the command through Follow Up Boss. A saved command is not proof of SMS delivery or completed provider cancellation.
AI calling — a partner request. With verified setup and an effective Execute permission, Ace writes the configured Reason for Calling field, confirms the saved reason, then applies the exact Leadengage outbound ai call tag. Leadengage performs the call. Trigger acceptance does not prove a call connected or completed.
Read the outcome. A draft, an accepted SMS command, a call trigger, a recorded handoff alert and pending cancellation are different states. Unknown outcomes require review and are not automatically replayed.
Three switches, in the order trust is earned.
- Enable Ace Trove on your Billing page (flat monthly tiers from $49/mo by database size). Autopilot is included — no separate SKU, no per-message fees from Ace.
- Open Autopilot in the admin dashboard, flip it on for the account, and enable the playbooks you want. Everything starts drafts-first.
- Graduate a playbook to autonomous sending once you've watched what it drafts — and take it back to drafts-first whenever you like.
Settings can be saved before account activation. Check the admin status and effective action modes to see what may run now. You also aim Autopilot yourself — at your whole eligible book, or only at contacts you tag ace-autopilot in Follow Up Boss (an ace-autopilot-exclude tag keeps it off any contact, in any mode). Nurture uses the same account switch, exclusions, review controls, and shared budget. Existing nurture tags work only after an admin explicitly adopts them for an enabled workflow in tagged mode. Autopilot workflows for new leads, re-engagement, and nurture are included with Ace Trove. Modern workflows do not add a per-nurtured-contact charge. SMS and calling use your own Lead Engage subscription and its provider charges. Moving to unified Autopilot does not automatically cancel or migrate an existing legacy nurture subscription; review any legacy billing separately.
What Autopilot is — and what it isn't.
Autopilot continues conversations only with verified current messages and authorized context. It can use available person-scoped messages, notes, history, deals, memory and relevant approved knowledge; it does not assume a complete thread or an unseen call recording. Restricted message bodies or unverified origins require review. Appointment preferences stay drafts: there is no calendar availability check, reservation or autonomous booking. It quotes no scores, probabilities, or equity figures to your leads, ever. Tag enrollment does not supply facts about a contact: no useful grounding means no message. Exclusions and opt-outs still win.
Availability depends on account activation, Ace Trove, saved permissions and current capability checks. Admin shows requested modes separately from the last verified effective modes. Recent activity is a bounded status view, not a complete history or a delivery count.
Your database is full of leads nobody has time to work. Now somebody does.
Ace Autopilot works them with your hands on the controls — drafts-first, explicitly enrolled, compliance-scanned, with verified follow-up conversations and qualified handoff to your agents. Included with Ace Trove. Self-serve, no sales call.