Skip to content

Free class · Systems and automation

How to build AI automations for your real estate business.

Every recurring job in your business is waiting on you to remember it. This class shows you how to write any of them as four decisions, so it runs on a condition instead of your memory. AI gathers, sorts, and drafts. You keep the judgment and the send button.

Ryan Wanner, founder of AI AccelerationInnovation Lab North, Compass Lakeside, HendersonvilleEvery class on this page is free to read and watch. No account needed.
Full class recording · 1 hr 7 min · recorded liveWatch on YouTube ↗

What you'll learn.

  • 01Why you're the trigger for every recurring job in your business, and why that's the bottleneck.
  • 02The four decisions behind every automation in any tool: trigger, middle, exit condition, and iterations.
  • 03The four kinds of trigger, and why every automation starts manual.
  • 04The research, analyze, report, act pattern that covers most recurring real estate work.
  • 05Why "it went rogue" is almost always a missing exit condition, and how to write one you can check.
  • 06The five rails, plus eight cookbook recipes that draft and never send.

Part 1

You are the trigger. That's the bottleneck.

The inbox you re-triage every morning. The market news you mean to keep up with. The ad numbers you check after the money's spent. Content that happens in bursts. The follow-up that falls off first in a busy week. None of these are hard. All of them are constant. That combination is exactly what a machine should be holding.

The cost isn't the minutes.

It's the re-deciding. You pay a startup tax every time, reloading where you left off and what good looks like. The standard drifts from Tuesday to Friday. It stops under pressure. And nothing that lives only in your head can be handed to anyone, person or machine.

What an automation is not.

It isn't autopilot. It does the gathering, sorting, and drafting while you keep the judgment. It isn't set-and-forget on day one. You watch it, fix the instructions, then stop watching. It isn't a new tool to buy. And it'll be wrong sometimes, which is why the rails exist.

The work you do every day isn't a to-do item. It's a specification you've never written down. Stop doing the task and start describing the conditions.

Part 2

Every automation is four decisions.

Miss one and it either never starts, never stops, or does the wrong thing forever. The tools will keep changing. These four won't.

  1. 01

    Trigger: what starts it

    A time, an event, a threshold, or you pressing the button. If nothing starts it, it isn't an automation.

  2. 02

    Middle: what it does

    The actual work, named step by step, in order. Research, analyze, report, and act only where you've allowed it. A vague middle produces a vague result.

  3. 03

    Exit condition: how it knows it's done

    The finish line, in writing, in checkable terms. "Stop when every email is in one of four buckets." Say what done looks like, not what good looks like.

  4. 04

    Iterations: how many tries

    One pass, or a capped number. "Try 3 times. If it still fails, stop and tell me exactly what broke." An automation that hides its failures is worse than none.

Four kinds of trigger. You already have all four.

Time

Every weekday at 6:00 AM. Every Monday. The most common and the most underused.

Event

A new email lands, a form is submitted, a lead comes in.

Threshold

Cost per lead passes your line, or a listing sits past its days-on-market mark.

Manual

You press the button. The right trigger while you're learning to trust it.

The middle that covers most of your work.

Research

Get the raw material: the ad numbers, the inbox, the news, the listings.

Analyze

Turn raw into meaning. What changed, what's off, what's noise. The step worth paying for.

Report

Short, ranked, with the recommendation stated. Something you can act on in a minute.

Act, only if allowed

Sort, label, draft, schedule, stage. Anything you haven't pre-approved waits as a draft.

Part 3

Write the spec, then turn one on.

Nobody opens a tool until four lines are filled in on paper: the trigger with a real time, the middle as named steps, an exit condition you could check, and a retry cap. The spec is the work. The tool is just where you paste it. Pick the one job that costs you the most. Everybody in the room built this one first.

The daily briefing

Every weekday at 6:00 AM, do this:

1. RESEARCH. Search for what changed in the last 24 hours in (a) AI tools that a real estate agent could actually use, and (b) the residential real estate market for [MY METRO].

2. ANALYZE. Drop anything that is a rumor, an ad, or a repeat of yesterday. Keep only items where something actually happened or a real number moved.

3. REPORT. Give me at most 5 items. For each one: one sentence on what happened, one sentence on why it matters to me specifically, and the source link. Rank them by what matters most to my business.

4. FLAG. Mark any number or claim I would need to verify before I repeat it to a client.

EXIT CONDITION: Stop when the briefing is under 250 words and has 5 items or fewer. If you find nothing worth reporting, say "Nothing worth your time today" and stop. Do not pad it.

ITERATIONS: If a search fails, retry up to 3 times, then tell me exactly what you could not reach.

Paste it into your assistant's scheduling feature and confirm it saved with a next-run time. Redact before you paste anything client-related. The second build in class was the inbox pass: every message sorted into draft, action, archive, or delete, replies written as drafts, and nothing sent.

The five-question gate.

  • Does it have a real trigger, with an actual time on it?
  • Is the middle written as named steps in order?
  • Is the exit condition something you could check, not argue about?
  • Can it tell you it found nothing, and tell you it failed?
  • Does anything reach a client without you pressing send? If yes, take it out.

The five rails.

  • Nothing sends to a client, a lead, or the public without a human pressing send.
  • No client names, addresses, phone numbers, emails, or financials go in. Redact to placeholders.
  • Every factual claim it hands you is flagged for verification.
  • It has a written exit condition and a capped number of retries.
  • It reports its failures out loud, in the same place it reports its wins.

Let it run for a week. The first time the output is wrong or too long, change one sentence and let it run again. The fix is almost always the exit condition. Once it's been right three days running, stop watching it and spec the next one.

Part 4

The Automation Cookbook.

The class came with a reference companion: a plain comparison of the assistants that can schedule AI work, and a set of recipes. One rule runs through all of it. Automate the preparation, not the execution. Have it draft, reconcile, or recommend, and keep sending, buying, deleting, and publishing behind your approval.

Pick one tool, not four.

Use the assistant you already pay for if it has a scheduling feature. Menu names and plan limits change, so when a page like this one disagrees with the product, trust the product. One automation actually running beats four accounts you opened once.

Build on what you own.

Your inbox, your calendar, your Drive, and your own sent mail. Portal leads arrive as email. Contract dates live in documents you have. Past-client history lives in what you've already written. Every recipe works from those.

Each recipe comes with its four decisions filled in, an approval gate that's never empty, and the legal line it stays behind. Every recipe drafts. None of them send.

The Unanswered Inquiry Sweep

Reads recent buyer and seller inquiries, finds every one you haven't answered, oldest first, and drafts a useful reply for each. Drafts only.

The Post-Close Loop

Checks your past-client sheet for anniversaries and seasonal windows in the next two weeks and drafts a touch with no ask in it.

The Deadline Watch

Reads your calendar and deal notes and ranks every inspection, appraisal, financing, title, and closing date in the next 72 hours, with the next action for each.

The Weekly Client Update

Drafts one update per active client: what happened, what's next with a date, what you need from them. If nothing moved, it says so. It never manufactures progress.

The Ad Loop

Pulls last week's campaign numbers, sorts every campaign into pause, scale, or test, and stages the changes without pushing anything live.

Daily Market and AI Research

The daily briefing with a stricter rule: primary sources only, a confidence label on every item, and nothing padded.

Listing and Market Content

Drafts content from real numbers and your notes, drops any number it can't source, and runs a Fair Housing pass that describes the property, never the people.

The Wire-Fraud Thread Check

Scans active-transaction email for lookalike sender domains, changed banking details, and new urgency. Read-only. It never confirms wiring instructions or marks anything safe.

Free with a student account

Get the workbook and prompts, free.

Create a free student portal account to get the workbook, prompts, templates, and downloads for every class.

Create my free account →

Already a student? The same link signs you in.

  • The Automation Spec Sheet
  • The automations prompt pack: daily briefing, inbox pass, ad loop, content, deep research
  • The Automation Cookbook: eight recipes, each with its approval gate
  • The class workbook with a seven-day plan
  • Automations skill card

Questions agents ask.

What is an AI automation, in plain terms?

A standing instruction instead of a one-off prompt. You write down what starts it, the steps in order, what done looks like, and how many times it retries before it tells you what broke. Then it runs on that condition instead of on your memory. It gathers, sorts, and drafts. You keep the judgment and the send button.

Which AI tool should I use to automate tasks?

Start with the assistant you already pay for, as long as it has a scheduling feature. ChatGPT, Claude, and Grok all have one. The menu is called something like Tasks, Scheduled Actions, or Routines, and the names move, so check your tool's own help pages. The four decisions work the same in every tool.

Why does my automation go off the rails or pad its report with filler?

Almost always a missing or vague exit condition. Write the finish line in checkable terms, like a word limit and a maximum number of items, and give it an explicit way to report that it found nothing. Cap the retries and make it report failures in the same place it reports wins.

Can an automation email or text my clients for me?

Not in this method. An automation can sort, label, draft, schedule, or stage. Nothing reaches a client, a lead, or the public until you press send. Commercial email and text messages carry their own legal rules, and a human check is the simplest way to stay inside them.

What should I automate first?

The recurring job that costs you the most, usually the inbox or keeping up with the news. The daily briefing is the easiest to turn on. Start it manually, fix one sentence each time it misses, and put it on a schedule once it's been right 3 times in a row. Then spec the next one.

Research sources

Check the current rules, capabilities, and source basis.

  1. 01OpenAI: A practical guide to building agents ↗

    First-party guidance on when agents fit, tools, instructions, guardrails, and human intervention.

  2. 02OpenAI: A business leader's guide to working with agents ↗

    First-party guidance on bounded autonomy, permissions, confirmation, human oversight, and audit trails.

  3. 03xAI: Grok Bot approvals, security, and privacy ↗

    Vendor documentation advising that automations prepare before they execute, with approval required for sending, purchasing, deleting, or publishing.

  4. 04NIST: AI Risk Management Framework ↗

    The framework behind bounding an automated system with explicit stop conditions and monitored failure reporting.

  5. 05FTC: CAN-SPAM Act compliance guide ↗

    The rules that govern any commercial email an automation drafts.

  6. 06NAR: Artificial Intelligence in Real Estate ↗

    NAR's current overview of AI use in real estate and policy priorities around privacy, Fair Housing, and copyrighted listing content.

Product access, limits, local listing rules, platform policies, and laws can change. Re-check the linked authority before important or public work.

Keep building

Put AI to work in your real estate business.

AI Acceleration: Real Estate is a private, self-directed curriculum for practical AI work across content, communication, lead follow-up, CRM, client research, operations, and relationship management. Build it. Review it. Keep what works.

Explore the Skool community ↗

Follow Ryan + AI Acceleration

No ranking, revenue, closing, or time-saving promise. The classes teach the work and the review systems around it.

Record: innovation-lab-public-classes. Approved record hash: add45e5ab0ee95b80ad70a4ecb50c8b8317c6997119c13fd8a727225815aff73. Claims: public-classes-free-to-watch, public-classes-portal-unlock, public-classes-taught-live.