CNAT TitleEducation Keeping AI on ThingsBook Time
Playbook Board
Level 3 · Autopilot · The graduation play

The Campaign Autopilot

One input (a listing, an event, a market shift) goes in, and a complete campaign comes out through the channels you already own.

Build over a week, run in an hour Two human checkpoints

The whole play, at a glance

Channel Mapdocument your tools, once
Triggerlisting, just-sold, market shift
Review + loadcheckpoint one
Approve + launchcheckpoint two
The resulta measured campaign that improves itself

You walk away with: campaign execution in one working session instead of a dozen: every asset drafted to fit your real channels, a load list, a send calendar, a measurement plan, and a debrief loop that makes the next campaign better than this one.

What you'll need

The foundation plays running: your Second Brain, and either a connected CRM or the CSV loop.

An honest list of your channels: which email tool, which social accounts, texting from where, landing pages built how. The autopilot can only fly the plane you actually own.

The setup

6 steps0 of 6 done

The output is a short document; save it in your workspace next to your Agent Profile. This is what lets the assistant produce assets in the right shape for each tool.

It lives in your workspace so every campaign starts from the same proven scaffold.

A new listing, a just-sold, a rate move, a seasonal push. Fill the kickoff brackets, run it, and review every asset in the package before anything gets loaded.

Campaign: Just Sold on Belmont › calendar
Day 1 · Sphere email"What the Belmont sale means for your street"Drafted
Day 3 · Social captionDemand story + market contextDrafted
Day 7 · Farm mailer copyTies to the landing page QRDrafted

The shape of the package: every asset drafted per channel, sequenced on a calendar, waiting on your approval.

All-in-one route: import the campaign into your CRM's campaign builder (the assistant formats content for copy-paste or CSV import to match). Multi-tool route: work down the load list the assistant produces, one channel at a time.

Sending always happens from your tools, under your name, with their unsubscribe handling.

Paste the numbers your tools show (opens, replies, clicks, form fills) into the Debrief prompt. Apply one improvement to the template. That edit is what makes this an autopilot instead of a one-off.

The prompts

The Channel MapRun once; becomes a reference document in your workspace
channel-map.prompt
Interview me about my marketing channels, one at a time: email (what tool, list size, how often I send), text (from what system, who's opted in), social (which platforms I actually post on), landing pages (how I build them), print/mail (if any).

Then write my CHANNEL MAP: a short reference document listing each channel, its tool, its audience, its format constraints (character limits, image sizes), and its role (nurture, announce, capture). I'll save this in my workspace so every future campaign is drafted to fit my real channels.
The Campaign KickoffTrigger in, full package out
campaign-kickoff.prompt
Using my Agent Profile and my Channel Map, build a complete campaign.

TRIGGER: [what's this campaign about: new listing / just sold / market update / event / seasonal]
DETAILS: [the facts: property, numbers with sources, dates]
AUDIENCE: [whole database / a segment / the farm]
GOAL: [conversations started / showings booked / home-value requests]
DURATION: [e.g. 2 weeks]

Produce the full package, organized by channel from my Channel Map:
1. Every asset, final-draft quality, in my voice, formatted to each channel's constraints.
2. A calendar: what goes out on which day, in which channel, in what order, and why that sequence.
3. A load list: the exact steps to get each asset into each tool, in order, so I can load the whole campaign in one sitting.
4. A measurement plan: the 3 numbers to check in two weeks and where I'll find them.

Fair-housing safe throughout. Flag any claim in my DETAILS that needs a source or a double-check before it ships.
The DebriefResults in, one compounding improvement out
debrief.prompt
The campaign ran. Here are the results from my tools: [paste opens, clicks, replies, form fills, conversations, and anything anecdotal]

Tell me: what worked, what underperformed, the single highest-impact change for next time, and rewrite the one weakest asset to show me the fix. Then give me the updated line to add to my Channel Map or Agent Profile so the improvement sticks.

Watch-outs

Two checkpoints, always. You review the full package before loading, and the schedule before launch. Automation earns trust; it never inherits it.

Consent lives in your tools. Email and especially text campaigns go only to people with proper opt-in, and your sending tools enforce unsubscribes. The assistant never sends directly.

One campaign at a time until the debrief loop works. The failure mode of automation isn't a bad email; it's ten mediocre campaigns nobody measured.

Advertising rules ride along: broker identification, license disclosures, and your brokerage's ad policies apply to every asset in the package.

Feeds this play

Trigger: buyer demandReverse Prospecting Engine Drop its demand story straight into the kickoff. Trigger: farm activityThe Farm Watchlist A record sale on the street is a campaign waiting to run.