“Create a 30-day social media calendar for a real estate agent.”

Thirty seconds later, the month is planned.

Monday is a motivational quote. Tuesday is a buyer tip. Wednesday is a market update. Thursday is a “dream home” post. Friday asks followers whether they prefer a modern kitchen or a traditional one.

The calendar is complete.

The marketing is not.

Nothing in those 30 posts could only have come from that agent. Nothing proves that the agent toured a house, advised a seller, studied a neighborhood, negotiated an inspection, or heard what buyers are worried about this week.

The content is clean. It is consistent. It is also interchangeable with the work of thousands of agents using the same prompt.

Agents make one central mistake with AI: they ask it to replace the source of their content instead of helping them extract, shape, and distribute what they actually know.

According to the National Association of REALTORS®’ 2025 Technology Survey, 75% of REALTORS® use social media, and 39% say it produces their highest number of quality leads. At the same time, 46% say AI has made no noticeable difference in their business.

That does not mean AI is ineffective. It means that producing more content and producing more business are different jobs.

Here are 11 ways agents use AI incorrectly, along with a better approach for each one.

1. You ask AI what to post

The wrong way

“Give me 20 social media ideas for a real estate agent.”

AI fills the empty space with the statistical average of real estate marketing:

• five buyer mistakes;

• three reasons to list now;

• a home-maintenance checklist;

• a mortgage-rate update;

• a “just sold” post;

• a quote about homeownership.

None of these subjects is inherently bad. The problem is that they begin inside the machine rather than inside the agent’s market.

AI has not attended your showings. It did not hear a buyer hesitate over the monthly payment. It did not see three sellers in the same subdivision make the same pricing mistake. It does not know why one house received four offers while the prettier one down the street is still sitting.

Without that information, AI fills the space with something generic.

The correct way

Stop asking, “What should I post?”

Ask:

• What happened in my business this week?

• What surprised a client?

• What objection did I hear more than once?

• What changed in my market?

• What did I notice while touring homes?

• What question did someone ask at an open house?

• What did I explain in a consultation that the public would benefit from hearing?

Record a three-minute voice memo after appointments, tours, inspections, and closings. At the end of the week, give those notes to AI and ask it to identify the strongest stories and lessons.

Your business should supply the raw material. AI should help turn it into content.

2. You make AI your strategist

The wrong way

Agents often ask AI to choose their audience, positioning, subject, format, hook, tone, and call to action in one prompt.

That looks efficient, but it hands every strategic decision to a system that does not know the agent’s business. The result may be a technically competent post aimed at no one in particular.

“Home buyers and sellers” is not an audience. It is most of the real estate market.

The correct way

Make the strategic decisions before AI starts writing.

Define:

1. Who is this for? First-time buyers competing below $500,000? Retirees deciding whether to sell a large home? Owners of downtown condos? Families relocating for a particular employer?

2. What decision are they facing? Are they deciding whether to wait, list, renovate, make an offer, change neighborhoods, or adjust their budget?

3. What do you know that could help? Bring a recent pattern, transaction lesson, market number, local change, or property comparison.

4. What should happen next? Should the reader save the post, answer a question, request a guide, attend an event, or schedule a conversation?

Then give AI the assignment.

AI can strengthen a strategy. It should not invent one at the same moment it writes the caption.

3. You use AI to imitate expertise

The wrong way

It is tempting to ask AI to sound authoritative:

“Write a thought-leadership post explaining why buyers should purchase now.”

The system can produce a confident argument. It may also invent a market trend, flatten uncertainty, generalize national data to a local market, or present an opinion as fact.

Confidence is cheap for a language model. Accountability is not.

The practical rule is simpler than the policy language: if your name is on the post, you own every sentence in it. “ChatGPT wrote that part” will not help when a buyer, seller, broker, or fellow agent asks where a claim came from.

The correct way

Use AI as a junior editor working from a verified evidence packet.

Before requesting the post, supply:

• the source of the market data;

• the period covered;

• the geographic area;

• the relevant comparison;

• confirmed listing facts;

• your interpretation;

• the limitations or uncertainty;

• the conclusion you are comfortable defending.

Then instruct AI:

“Use only the supplied facts. Do not add predictions, guarantees, property details, neighborhood claims, or statistics. If information is missing, flag it rather than filling the gap.”

AI should improve the explanation of your expertise. It should never manufacture expertise for you.

4. You let AI speak in its own voice

The wrong way

You recognize this caption immediately:

“Whether you’re dreaming of your first home, considering your next move, or simply curious about the market, I’m here to guide you every step of the way. Let’s turn your real estate dreams into reality!”

It is warm, grammatical, polished, and almost completely empty.

The warning signs are familiar:

• “in today’s ever-changing market”;

• “now more than ever”;

• “unlock the door”;

• “more than just a house”;

• “your real estate journey”;

• “nestled in the heart of”;

• “let’s connect”;

• “dream home”;

• excessive em dashes;

• a row of decorative emojis;

• a conclusion that summarizes what the reader just read.

The danger is not that people will accuse you of using AI. The danger is that they will learn nothing about you.

The correct way

Build a simple voice guide containing:

• how you speak with clients;

• three to five examples of your real writing;

• phrases you regularly use;

• phrases you never use;

• your typical sentence length;

• how direct, humorous, analytical, or conversational you are;

• opinions you genuinely hold;

• claims you will not make.

Then make one final edit with your ears, rather than your eyes.

Read the post aloud. If you would never say a sentence to a client, rewrite it in the words you would use.

AI can supply the skeleton. Your language should remain on the surface.

5. You use AI to create more instead of extract more

The wrong way

An agent uses AI to generate five unrelated posts every week.

This creates an endless appetite for new subjects. The agent is always feeding the calendar and rarely develops an idea deeply enough to become known for it.

More posts do not necessarily create more attention. Meta says it may reduce the distribution of accounts that repeatedly publish low-quality, repetitive, engagement-bait, or unoriginal content.

Volume without substance can accelerate invisibility.

The correct way

Use AI to extract more value from one strong piece of original knowledge.

Suppose you tour a house that looks renovated online but reveals expensive compromises in person. That one experience could become:

• a Reel: “Three renovations that photograph better than they perform”;

• a Facebook post explaining what buyers missed;

• a carousel showing what to inspect;

• an Instagram Story poll;

• a seller post about which upgrades buyers actually notice;

• a buyer email;

• an open-house checklist;

• responses to common questions in the comments.

One original observation becomes eight useful assets.

Explore one valuable subject from several angles. Do not make one post appear eight times.

6. You create content at your desk instead of documenting your work

The wrong way

At the end of the month, the agent sits in front of a laptop trying to “create content.”

That is difficult because the most compelling material has already disappeared:

• the expression on a buyer’s face when the backyard opens into view;

• the noise that did not appear in the listing photos;

• the afternoon light in the living room;

• the inspection item everyone initially underestimated;

• the conversation that changed a seller’s pricing decision;

• the difference between two streets in the same neighborhood.

AI cannot recover moments the agent never captured.

The correct way

Think like a documentarian before thinking like a marketer.

While working, collect:

• ten-second video clips;

• photos of non-confidential details;

• voice notes;

• recurring questions;

• screenshots of approved data;

• observations about layout, land, light, traffic, maintenance, and buyer response;

• anonymized transaction lessons.

Luxury broker Latham Jenkins describes his content as an extension of the work he already does: walking land, studying light, watching weather, and noticing how a place functions. One of his property videos reached millions of views because it conveyed the experience of being there, rather than repeating the feature list.

Do the real work. Capture the real work. Let AI help shape the record.

7. You use adjectives where you need local evidence

The wrong way

AI is excellent at producing polished vagueness:

• highly desirable neighborhood;

• vibrant community;

• ideal location;

• sought-after area;

• perfect for families;

• convenient to everything;

• impressive appreciation;

• top-rated schools;

• safe and quiet community.

Some of these claims are merely forgettable. Others can get awkward fast. “Family-friendly,” “safe,” and coded descriptions of who belongs in an area tell the reader very little about the property, and they can wander into Fair Housing territory.

A better habit is to describe what a person can actually see, measure, or verify, then let the reader decide whether the place suits them.

The correct way

Replace characterization with observable detail.

Instead of:

“Located in a vibrant, family-friendly neighborhood.”

Write:

“The house is 0.4 miles from Riverside Park, two blocks from the Main Street bus stop, and inside the attendance boundary listed by the district for Edison Elementary. Verify enrollment and boundaries directly with the district.”

Instead of:

“A highly desirable area with strong appreciation.”

Write:

“The neighborhood’s median sale price increased from $ to $ between these two periods, according to ___. Individual properties vary.”

Instead of:

“Perfect for a young family.”

Describe the fenced yard, bedroom configuration, first-floor layout, nearby named amenities, measured distances, transit access, and objective property features.

Describe the home and the place. Let the consumer decide whether it fits.

AI becomes more useful when you give it harder facts and fewer adjectives.

8. You let AI create a property that does not exist

The wrong way

An agent uses generative tools to:

• replace a cloudy sky;

• widen a room;

• add a view through a window;

• remove power lines;

• erase a water stain;

• furnish an empty room;

• change the landscaping;

• turn a still photograph into synthetic video;

• create a dramatic renovation that has not happened.

The content looks impressive. The trouble starts when the improved image gives a buyer the wrong idea about what is really there.

You do not need to memorize a stack of rules before touching an editing tool. Start with common sense: do not make the room larger, the view better, the condition cleaner, or the surroundings more attractive than they are. Then check your brokerage and MLS rules before publishing edited listing media. Requirements can vary, especially when an image has been materially altered.

The correct way

Use a simple test:

Could this alteration influence a reasonable buyer’s understanding of the property’s condition, features, dimensions, layout, view, finishes, or surroundings?

If the answer is yes, pause before posting. Find out whether the edit is allowed, label it clearly when disclosure is required, keep the original, and never use AI to hide a defect or invent a material feature.

Virtual staging can help people imagine a space. It should not ask them to imagine that the space itself is different.

When in doubt, use AI to plan a shot list, write the video script, create captions, remove background noise, or organize footage. Do not use it to rewrite physical reality.

9. You copy one AI post onto every platform

The wrong way

The same caption, aspect ratio, opening, hashtags, and call to action appear on Facebook, Instagram, LinkedIn, TikTok, YouTube Shorts, and every local group.

That is duplication, rather than repurposing.

People behave differently on different surfaces:

• Reels must earn attention immediately.

• Stories can be casual and interactive.

• Facebook posts can hold a longer local story.

• Community groups expect relevance and participation, rather than drive-by promotion.

• Carousels need a visual progression.

• Email allows more context.

• A short video shown to non-followers must make sense without prior knowledge of the agent.

Instagram also says it prioritizes original content and may replace identical reposted material with the original in recommendation surfaces.

The correct way

Preserve the central idea, but change the delivery.

For one market observation, ask AI to create:

• a three-second video hook;

• a 35-second spoken script;

• a five-slide carousel;

• a Facebook story with local context;

• a group post that invites useful community experience;

• an email explanation;

• a direct-message follow-up.

Then review each version in the environment where it will appear.

Repurpose the insight, rather than merely copying the file.

10. You optimize for reactions instead of conversations

The wrong way

AI is told to “make this more engaging,” so it adds:

• “Double-tap if you agree.”

• “Tag someone who needs to see this.”

• “Comment YES if you want to buy a home.”

• “Like, share, and follow for more.”

• “Which kitchen would you choose: A or B?”

These prompts can produce activity without intent. Meta explicitly identifies repetitive engagement bait as low-quality content that may be demoted.

A hundred people choosing Kitchen A does not mean one of them needs an agent.

The correct way

Invite a response that reveals a real question, decision, or need.

For example:

• “What is the one cost you still cannot estimate confidently?”

• “Would more inventory make you start looking, or is the monthly payment still the deciding factor?”

• “Which two neighborhoods are you comparing?”

• “If you are considering selling this year, what would need to happen first?”

• “Want the five comparable sales behind this number? Message me COMPS and I’ll send them.”

• “I have two similar homes that are not in this post. If this layout is what you need, ask me for the list.”

The offered resource must exist and be delivered promptly. Do not use a promised guide as a pretext for an unrelated sales sequence.

The purpose of content is to open a useful conversation.

11. You automate publishing but not listening

The wrong way

The agent builds an efficient machine:

1. AI writes the posts.

2. A design tool builds the graphics.

3. A scheduler publishes them.

4. The content calendar turns green.

5. Comments and messages sit unanswered.

6. No one records which subjects created conversations.

7. The next month begins with another generic calendar.

The system has automated the least valuable part and neglected the work closest to revenue.

The correct way

Build the workflow around listening and follow-up.

After every post:

• respond to substantive comments;

• continue appropriate conversations in DM;

• deliver promised resources;

• record genuine inquiries in the CRM;

• tag the subject and source;

• note the questions people asked;

• create follow-up content from those questions;

• call when the person explicitly requests help;

• track what became an appointment.

AI can help by summarizing comments, grouping recurring questions, preparing draft replies, extracting lead context, producing a brief before the agent calls, reminding the agent about unfinished conversations, and identifying which subjects attracted the right people.

The human relationship should remain human.

Automating publication creates output. Listening creates insight. Following up creates business.

The role AI should play

AI should become the operating system around your expertise, rather than a substitute personality.

It can listen to a voice memo and find the story inside it. It can turn one market observation into several formats. It can simplify an explanation, organize a content library, identify recurring questions, draft responses, and keep good ideas from disappearing.

The agents who benefit most from AI will give it rich human material and refuse to publish anything that no longer sounds, looks, or feels true.

AI moves fast. Your experience gives it somewhere worth going.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A practical guide for operators building with AI.

Local agents post what they are seeing in negotiations. Builders announce incentives. Mortgage professionals react to rate moves. Economists share fresh charts. Reporters flag new developments, and residents surface neighborhood stories long before they become polished market reports.

The information arrives every morning in pieces. By the time you have opened ten tabs, checked the date on everything, and worked out which claims deserve attention, the morning is gone.

This is where Grok 4.5 changes the workflow. It is the only frontier model with native live access to the X firehose through xAI’s X Search tooling. Grok can search the conversation as it unfolds, follow threads, find relevant accounts, and combine those posts with broader web research. Instead of scrolling until something useful appears, you can ask for a focused market brief built around your city, property segment, and clients.

You can do that manually on the web in a few minutes, or give the workflow to an AI agent and have the finished brief delivered on a schedule.

Why Grok 4.5 and X work well together

Real estate information moves at several speeds. Official market reports arrive on a schedule. MLS statistics describe transactions that have already occurred. Local government pages publish agendas, permits, and planning documents. X captures the reaction in between those releases.

That makes X useful for more than social media marketing. It is a live research source where agents, economists, builders, reporters, investors, lenders, and residents reveal what they are seeing and watching. You can spot a pricing concern, incentive change, policy debate, or neighborhood story before it reaches a monthly presentation.

Most frontier models can search the public web. Grok 4.5 has something different: native live access to the X firehose through xAI’s X Search tooling. It can perform keyword and semantic searches, focus on selected accounts, restrict results by date, retrieve complete threads, and inspect media. Its Web Search tool can then browse current pages outside X, and both tools can return citations.

It can find the live conversation on X, trace an important claim to its source, compare competing views, and organize the result around your market and clients.

This is what you can do today:

Use X to discover what deserves a closer look. Use MLS reports, Realtor association releases, government pages, lender data, company announcements, and credible local reporting to confirm the facts you repeat to clients.

Approach 1: build the brief on the Grok website

This is the fastest place to begin because you can improve the briefing before you automate it.

Open Grok on the web or inside X. If Grok 4.5 appears in your model picker, select it. Consumer access can vary by plan and region, so use the strongest Grok model available.

Start with one market and one time window. Broad prompts such as “Tell me about real estate” produce broad answers. Give Grok the geography, property segment, sources, and output you actually need.

Try this prompt:

The first result will teach you what the prompt is missing. You may need to add your counties, price bands, school districts, builder names, or local data sources. You can also give Grok a short account list and ask it to prioritize those voices while still looking for contrary evidence.

A useful follow-up is:

That second pass is where the brief becomes more trustworthy. It prevents a lively X conversation from being mistaken for a measured change in the market.

For a weekly version, widen the time window and ask for movement rather than repetition:

After a few runs, save the prompt that consistently gives you a clean result. You now have the specification for an automated agent.

Approach 2: have an AI agent deliver the brief

The manual workflow still asks you to remember to run it. An AI agent can run the same research every morning, keep the format consistent, compare today’s findings with prior briefs, and send the result to Telegram, email, Slack, or another channel.

Hermes Agent is one option. It is an open-source agent that can use Grok as its model, connect to research tools, run scheduled jobs in fresh sessions, and deliver the final response to a messaging channel. Other agent frameworks can follow the same design.

Step 1: install Hermes and sign in to xAI

Install Hermes Agent, then complete its guided setup:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
hermes setup

The simplest X Search credential is an existing SuperGrok or X Premium+ subscription. Hermes can use it through xAI OAuth, so you do not have to create, copy, or store an API key:

hermes auth add xai-oauth

Hermes uses xAI’s device-code login. It opens or prints a verification URL and code; sign in and approve access in any browser. There is no local callback requirement, so the browser can be on another computer or phone. For a remote server, use hermes auth add xai-oauth --no-browser and open the printed URL yourself.

Confirm that Hermes can refresh the credential:

hermes doctor

Look for xAI OAuth: logged in under Auth Providers. xAI controls subscription entitlements, so a browser login can succeed while the API later returns 403. If that happens, use a paid XAI_API_KEY or an eligible SuperGrok/X Premium+ tier.

Step 2: select Grok 4.5 and test ordinary chat

Run the model picker:

hermes model

Select xAI Grok OAuth (SuperGrok / X Premium+), then choose grok-4.5. Start Hermes and ask it to reply with a short phrase. This tests the model connection before X Search adds another moving part.

Step 3: enable Hermes’s native X Search tool

Current Hermes releases include a dedicated x_search tool. It calls xAI’s server-side X Search through https://api.x.ai/v1/responses and returns a synthesized answer with citations to the originating X posts. For read-only market research, you do not need an X Developer App, xurl, or a separate MCP server.

OAuth normally enables the tool automatically. Check it:

hermes tools list

You should see x_search or X (Twitter) Search marked enabled. If it is disabled, run hermes tools, open the Search category, and enable X Search.

Pin Grok 4.5 as the model Hermes uses for the server-side search call:

hermes config set x_search.model grok-4.5

Step 4: prove X Search works before scheduling it

Open Hermes and run a narrow test with a known active account and a recent date range:

A passing result includes direct x.com/.../status/... citations. If Hermes returns an answer without citations, do not count it as a successful X search. Narrow handle or date filters can produce a degraded result in which Grok answers from model knowledge because the X index found no matching posts.

Use web_search separately to verify consequential claims against an MLS report, government page, company announcement, lender source, or credible local newsroom. X Search tells you what people are saying; web research helps establish what is true.

Step 5: write the briefing contract

Scheduled agents start each run in a fresh session, so the instruction needs to stand on its own. Give the agent a fixed research contract rather than a vague request to “send real estate news.”

Use this as a starting prompt:

Add a short list of preferred local accounts and domains. Keep it selective. Twenty reliable sources will usually beat two hundred loosely related accounts.

Step 6: schedule and deliver it

Hermes can create a scheduled job through natural conversation. Once your Telegram, email, Slack, or other destination is connected, tell the agent:

For a weekly briefing, schedule a separate job:

Hermes runs scheduled tasks in fresh sessions and can attach the research skills needed for the job. A continuable delivery lets you reply to the brief and ask, “Which of these stories deserves a client email?” without having to paste the report back into the conversation.

Step 7: test before you trust

Run the briefing manually for a week before treating it as part of your operating rhythm.

Check whether every link opens. Look for old stories presented as new. Confirm that market statistics refer to the correct geography and reporting period. Watch for duplicated commentary that traces back to one original post. Make sure rumors stay in the watchlist. If the agent missed a local source you rely on, add it to the contract.

Keep a human approval step before publishing client-facing content. The automated brief is an internal research product. It should help you decide what to investigate, discuss, or write. It should not publish rumors, quote private client information, or turn uncertain market chatter into advice under your name.

What a useful brief changes

The goal is not another newsletter in your inbox. The goal is a better first hour of the day.

A good brief may tell you that a builder increased incentives in two nearby communities, a proposed zoning change reached the council agenda, local agents are reporting longer negotiations on entry-level homes, and the latest association report confirms rising inventory. You can then decide which item deserves a call to a lender, a message to a buyer, or a short client email.

It also creates a content advantage. Most agents wait for a monthly report, copy a chart, and post the same summary as everyone else. A disciplined daily briefer helps you see the question forming before the polished report arrives. You can explain what people are noticing, what the verified data says, and where the two still disagree.

That is a more useful kind of local authority. It comes from showing your work, keeping uncertainty visible, and connecting a market development to an actual client decision.

Try this before next week

Run the web prompt for your market on three consecutive mornings. Save each brief. On the third day, ask Grok to compare all three and identify which signals strengthened, weakened, or vanished. If the result improves a real client conversation, turn the prompt into a scheduled agent job.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.

They may call it a system, a playbook, a business operating model, or simply “the way we do things here.” The language changes. The point does not. If you want a real business instead of a personal production treadmill, the work has to become repeatable.

Some coaches will tell you the number of precesses you need is over a dozen. They may be underselling it.

I agree with the premise. Processes raise revenue when the team executes them well. They reduce randomness. They keep the best parts of the business from living inside one agent’s head. They make training easier. They make delegation safer. They make growth possible without turning every week into a rescue mission.

A good process is leverage.

The uncomfortable part is that the process itself rarely fixes the business.

The process binder does not do the work

Most agents do not fail because they hate systems.

They fail because real estate is a constant interruption machine.

A lead comes in while you are driving. A seller texts about showings. A buyer wants to see a house tonight. An inspection report lands. A lender needs one more document. A past client asks for a contractor. Your content calendar says to post a local market update, but the hot lead in your inbox feels more urgent.

By 4 p.m., the beautiful Monday morning process has been negotiated down into whatever you remembered to do.

That is human nature. We are bad at perfect repetition, especially when the work is emotional, fragmented, and full of context switching. We skip the follow-up step because the client sounded fine. We forget to update the CRM because we already know what happened. We rewrite the same email because finding the template feels slower. We delay the seller update because there was nothing exciting to say. We promise ourselves we will catch up later.

Later becomes Friday.

The process did not fail. The human execution failed.

That distinction matters because it changes the solution. The answer is not another PDF checklist, another training video, or another dashboard nobody opens after week two. Agents need an execution solution that can carry the process when humans are busy being human.

That is where AI agents become practical.

AI agents are built for the boring part

For an AI agent, repetition is not annoying. It does not get bored with checklists. It does not decide the follow-up step is probably fine to skip because the client sounded happy. It does not forget the template because a new deal came in.

That makes AI agents useful in a process-heavy business.

A weekly seller update is a good example. A chatbot can help write one if you remember to prompt it. An agent can look for listings that need updates, pull the seller-update process, gather showing feedback, draft the message, flag missing MLS activity, and hand you the draft before the promised send time.

The same idea applies to listing launches, buyer consultations, open house follow-up, and contract-to-close work.

One clever caption is not the prize. The prize is making the business behave the way the owner already says it should behave.

AI Agents are useful here because the work is procedural. Sticking to processes feels unnatural for people. It is normal for software.

Where Hermes or OpenClaw fits

You need an AI agent to operate the process. I have used Hermes and OpenClaw for this kind of work, but you do not need to start with both.

Pick one.

Hermes is a strong choice if you want an agent you can use from terminal, Telegram, and local tools. It can read files, run scripts, schedule jobs, call MCP servers, and keep reusable skills. For a real estate operator, Hermes can become the assistant that checks the process and prepares the next action.

OpenClaw is a strong choice if you want a broader agent gateway and execution environment. It can run agent work through its own gateway, channels, and task systems. If your setup is already built around OpenClaw, use that as the agent layer.

The important part is not the brand name. The AI agent needs access to your process documents and a safe way to draft, check, and ask for approval.

GBrain is the memory layer underneath that agent.

What GBrain does

GBrain is built by Garry Tan, President and CEO of Y Combinator. It gives the AI agent a memory of how your business actually works.

Think about the question every agent has asked before a call: “What do I need to know before this meeting?” A search engine can find notes. A brain can read the notes and hand back the answer.

For example, imagine you have a listing meeting tomorrow with a past client named Maria. You ask the agent:

What do I need to know before my meeting with Maria tomorrow?

GBrain can pull the source pages behind that relationship and return something like this:

Maria bought with you in 2019 and referred her sister in 2022. She mentioned in your last call that she may sell if her commute gets worse. Three things are still open:

1. She asked for a valuation range after the spring comps came in.
2. You promised to send her the contractor list for light kitchen updates.
3. Her daughter is starting school in August, so timing may be sensitive.

Heads up: nothing has been added to the brain about Maria since March. If she replied by text or called your personal phone, the brain may not have that context yet. Ask a catch-up question before assuming this is current.

That last warning is a big deal. A useful brain tells you what it knows and what it might be missing, so you can walk into the meeting prepared without acting overconfident.

The same pattern works for process. If a seller asks why traffic has slowed down after two quiet weekends, the agent can pull your seller-update process, the original listing-positioning notes, last week’s showing feedback, the pricing conversation script your team approved, and the neighborhood market notes you saved after the last team meeting. The draft it prepares sounds less generic because it is using your business as context.

The technology underneath is simple enough: GBrain stores your notes as searchable pages, breaks long documents into chunks, and uses embeddings so the agent can find related ideas even when the wording is different. One file might say “weekly seller update.” Another might say “Friday showing recap.” GBrain helps the agent understand those are probably part of the same workflow.

That is the difference between a folder of documents and a working brain. Search finds the pages. GBrain helps the agent read them, connect them, cite them, and notice the gaps before the work reaches a client.

The simple setup

This is the technical version of the stack. You can run it on a Mac mini, a small server, or a VPS. If you are a solo agent, start small. If you run a team, treat this like an internal operating system project.

Step 1: Install an AI agent

Choose either Hermes or OpenClaw.

If you choose Hermes, start here:

https://hermes-agent.nousresearch.com/docs

A basic Hermes install looks like this:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
hermes setup
hermes doctor

If you choose OpenClaw, start here:

https://docs.openclaw.ai/install

Then run its setup and health check:

openclaw setup
openclaw doctor --fix
openclaw status

Pick the one you are actually going to use. Do not make this a weekend infrastructure project.

Step 2: Ask the agent to install GBrain

Send your agent this exact message:

Retrieve and follow the instructions at:
https://raw.githubusercontent.com/garrytan/gbrain/master/INSTALL_FOR_AGENTS.md

That install guide is written for agents. It tells the agent to install GBrain, initialize the brain, choose the search mode, import files, and verify the setup.

If the agent asks where your process files live, start with one folder. Something like:

~/agency-brain/

Inside that folder, keep a few Markdown files:

listing-launch.md
weekly-seller-update.md
lead-follow-up.md
contract-to-close.md

Do not dump your whole Google Drive into the brain on the first pass. Give it clean process files first.

Step 3: Set up embeddings

GBrain can work with keyword search, but embeddings make it much more useful.

Use an embedding provider such as ZeroEntropy or OpenAI. Add the API key when the installer asks for it. Then have the agent verify the setup with:

gbrain doctor --json
gbrain stats

The outcome you want is simple:

• GBrain is installed.

• Your process files are imported.

• Embeddings are working.

• Your AI agent can search GBrain before drafting or checking process work.

That is enough for a real first version.

Four high-value use cases

1. Client meeting prep

This is the clearest way to feel the difference between search and a brain.

Before a listing appointment, buyer consultation, price-reduction call, or past-client check-in, ask:

What do I need to know before my meeting with [client name] tomorrow?

From GBrain, the agent can assemble a clean brief from prior calls, old promises, valuation notes, showing feedback, family timing, referral history, and open follow-up items. A good answer should include sources and a warning when the brain may be stale. If the client has been texting you outside the system, the agent should say so instead of pretending it sees everything.

That turns meeting prep from a scavenger hunt into a five-minute review.

2. Weekly seller updates

This is the first recurring process I would automate because it is valuable and often skipped.

For weekly seller updates, the agent pulls the process from GBrain, gathers showing feedback and market activity, drafts the update, and flags missing information. The listing agent still approves the recommendation and tone before anything goes out.

A useful first prompt:

Search GBrain for our weekly seller update process. For each active listing, prepare a draft update using the required sections. Flag missing showing feedback, MLS activity, pricing discussion points, or seller approval items. Do not send anything. Return drafts for review.

3. Listing launch checklist

A listing launch has too many details to trust memory.

For a launch, use the agent as the checklist runner. It compares the process against the actual file: property facts, photo notes, MLS copy, seller positioning, email announcement, social captions, open house plan, neighbor outreach, and compliance review. It does not have to publish anything. Its first job is to show what is complete, what is missing, and what needs human approval.

This alone can save a team from the tiny misses that make a listing feel sloppy.

4. Contract-to-close checklist

Transaction coordination rewards boring consistency.

In contract-to-close, the same habit becomes deadline protection. The agent compares the file against your process: executed contract, earnest money, option period, inspection, lender milestones, appraisal, title, HOA docs, repairs, closing disclosure, utility reminders, walkthrough, keys, and review request.

A human coordinator still handles judgment calls and relationships. The agent watches the checklist and raises its hand early.

The business becomes a loop

The best version of this system is simple.

You write down the process. GBrain stores it. The agent searches the brain when work begins. The agent drafts or checks the repeatable parts. A human approves anything client-facing or judgment-heavy. After the work is done, the lesson goes back into GBrain.

That last step is easy to miss. It is also where the system gets better.

A bad seller conversation can become a better pricing process. A missed inspection deadline can become a stronger transaction checklist. A successful listing launch can become a reusable example. A client complaint can become a process update instead of a memory that fades after the next emergency.

Most teams lose those lessons. They fix the problem, move on, and meet the same problem again two months later.

GBrain gives the lesson a place to live. The agent gives the lesson a way to return when the next workflow starts.

Often take

Real estate teams already know they need processes. The harder question is how to keep those processes alive during a real week.

That is why AI agents are more interesting than another prompt pack. A prompt helps when someone remembers to use it. An agent connected to GBrain can pull the playbook at the moment of work, check the missing pieces, prepare the draft, and bring the human back where judgment is needed.

The biggest gain is consistency. A team can launch listings with the same standard, update sellers with the same rhythm, follow up with leads through the same sequence, and carry transaction lessons into the next file.

Real estate coaches were right about process. What changed is that the execution layer finally exists.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.

The onboarding process can feel like walking into a city where every building has a different login screen. There is Passport, My eXp, eXp World, Workplace, eXp University, eXpert Care, SkySlope, etc, etc. Every time you think you found the right place, someone tells you, “Oh, that moved,” or “That depends on your state,” or “Ask your broker team,” or “Search for it in Workplace.”

I feel you. I used to be a fellow eXp agent, and I remember that first stretch clearly. eXp has a huge amount of support, training, and software, but the hard part for a new agent is knowing what exists, where to find it, and what order to do things in. Most agents get stuck because the path is scattered across too many portals, too many acronyms, and too many “ask this department” answers.

So I decided to use a little AI magic. I built a skill called eXp Agent Onboarding Coach, and the goal is simple: give new eXp agents a calm, practical guide through the first days and weeks after joining.

The skill turns your AI assistant into an onboarding coach for eXp agents. You can ask, “I just joined eXp. What should I do first?” and it will give you a staged plan instead of generic motivation.

It helps you figure out what to do today, what to verify with your broker or country team, which eXp tools to log into first, which training to watch, how to approach CRM setup, what to avoid before your license is active, what to ask your mentor, and what to prepare before your first transaction.

Use the skill as a support layer alongside eXp support, your broker, and your mentor. Think of it as the organized friend sitting next to you, sorting the mess into a checklist: do this first, ignore that for now, ask this person before you act, and save this link because you will need it again.

I built the skill around three practical jobs. First, it gives new agents a clearer onboarding map. Second, it helps agents discover eXp tools they may not know they already have. Third, it keeps reminding them when something needs official verification from the right source. I built it with a global view from the beginning, because eXp is not a one-market brokerage. The skill is designed to ask where the agent works before it gets specific, so a California agent, a Texas agent, an Ontario agent, a UK agent, and a South Africa agent do not get treated as if they are operating under the same rules.

How to install and use it

You need two things: an AI agent and the eXp Agent Onboarding Coach skill.

The simplest technical setup is Hermes Agent, but the skill also works with Claude Projects, ChatGPT custom GPTs, Cursor, Claude Code, OpenClaw, and most other AI agent tools that can read Markdown instruction files.

Hermes is an open-source AI agent from Nous Research. It can install skills, remember workflows, use tools, and run from your computer.

To install Hermes, open your terminal and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

After that, run the setup wizard:

hermes setup

During setup, Hermes will ask you to choose a model provider. You can use OpenRouter, Anthropic, OpenAI, Gemini, local models, or another supported provider. Once Hermes is ready, install the eXp Agent Onboarding Coach skill with this command:

hermes skills install https://raw.githubusercontent.com/oftento/exp-agent-onboarding-coach/main/SKILL.md --category productivity

Then start Hermes:

hermes

Load the skill inside Hermes:

/skill exp-agent-onboarding-coach

Now you can ask your onboarding question in plain English. For example:

I just joined eXp Realty. I am active in Texas, but I am overwhelmed. I have Passport access but I have not set up CRM or MLS yet. What should I do first?

If you prefer Claude, download the skill files from GitHub , then add SKILL.md, CLAUDE.md, AGENTS.md, the references folder, and the templates folder to a Claude Project. Start with this prompt: “Use the eXp Agent Onboarding Coach instructions. I just joined eXp. First ask only the questions you need, then give me today’s plan and this week’s checklist.”

If you use ChatGPT or a custom GPT, download the latest release ZIP, upload the files into a custom GPT knowledge base, or paste the SKILL.md instructions into your GPT configuration. Use a short instruction like this: “You are an eXp Agent Onboarding Coach. Follow the uploaded SKILL.md. Always ask for the agent’s country/region and local market before giving detailed onboarding steps. End every answer with concrete next actions.”

For Cursor, Claude Code, OpenClaw, or another agent tool, put SKILL.md, AGENTS.md, CLAUDE.md, references/, and templates/ in your project folder. Then tell the agent: “Follow AGENTS.md and SKILL.md. Act as an eXp Agent Onboarding Coach.” The skill is plain Markdown, so it travels well across systems.

The GitHub repo

The skill is public here: https://github.com/oftento/exp-agent-onboarding-coach.

The latest release is here: https://github.com/oftento/exp-agent-onboarding-coach/releases/latest.

The direct Hermes install URL is https://raw.githubusercontent.com/oftento/exp-agent-onboarding-coach/main/SKILL.md.

Once it is installed

Start with a simple prompt. You can copy this:

I just joined eXp Realty.

Country/region:
State/province/market:
My status: application submitted / waiting for activation / active / not sure
Do I have Passport access? yes/no
Do I have MLS/board/portal access? yes/no/not applicable
Am I brand new or transferring from another brokerage?

Please give me:
1. what I should do today;
2. what I should do this week;
3. what tools I should log into first;
4. what I should ask my broker/mentor/country team;
5. what I should not do yet.

Here is the kind of answer you might get:

## Your current stage

You still have two operational gaps: MLS/board/portal access and CRM setup.

## Today’s plan

- [ ] Confirm your license/affiliation is active under eXp in your market.
- [ ] Log into Passport / Okta.
- [ ] Confirm access to My eXp.
- [ ] Find eXpert Care in eXp World.
- [ ] Join your local Workplace groups.
- [ ] Create an eXp Support Directory.
- [ ] Ask your local broker/team what must be completed before advertising or working with clients.

## Tools to check first

1. Passport / Okta
2. My eXp
3. Workplace
4. eXp University
5. eXp World / eXpert Care
6. CRM of Choice: BoldTrail, Cloze, Lofty
7. SkySlope or your local transaction platform

## CRM note

Before paying for an outside CRM, check eXp CRM of Choice. BoldTrail, Cloze, and Lofty may be included for standard use through your existing eXp tech package. Verify current availability, IDX/MLS fees, add-ons, and country/market rules.

## Ask your broker or local team

- Am I cleared to publicly announce I am with eXp?
- What exact brokerage name and license disclosures must I use?
- Which MLS/board/portal steps are still pending?
- Which transaction platform should I learn first?
- Am I assigned a mentor?

## Do not do yet

- Do not advertise yourself as active until your affiliation is confirmed.
- Do not publish ads without checking required disclosures.
- Do not bulk text or email your database without consent/compliance review.
- Do not buy leads before your CRM and follow-up system are ready.

That is the whole point. The skill gives you a checklist, but it also gives you timing. New agents need the right next step, the right support contact, and the right warning before they accidentally publish something or buy something they should have checked first.

Why this matters for recruiters and sponsors

If you are an eXp recruiter, sponsor, team lead, or mentor, this is something you can send to your downline as a support layer. A new agent often asks the same questions over and over: Where do I log in? What is Passport? Do I use My eXp or Enterprise? Where do I find training? What CRM should I use? Wait, I can use Cloze and Lofty too? Who do I ask about MLS? Can I post my announcement yet? What do I do this week?

This skill gives them a starting point before they panic, guess, or buy tools they may not need. If it helps one agent avoid three days of confusion, it is worth sharing. If it helps a whole downline get oriented faster, even better.

What we should tackle next

Often is built around a simple promise for working real estate agents: AI moves fast. Agents move with Often. The Often Letter is the weekly version of that promise. The goal is less hype and more useful work. Each issue should help an agent answer one practical question: what should I understand, try, automate, delegate, or change in my business this week because of AI?

That means we care less about abstract AI news and more about the work sitting on an agent’s desk right now. When we cover a tool, we want to show what happened, why it matters to an agent’s business, and what to do next. When we build a skill like this one, we want it to become a practical field guide, something an agent can open when they are stuck and use before lunch.

So yes, share this with any eXp agent who is stuck in onboarding. Send it to your downline, your team, your new agents, your mentor group, or the person who just texted you because they cannot find the right portal.

But I also want the conversation to be broader than this one skill. Tell me where the friction is in your business. Is it follow-up? Listing prep? Social content? Recruiting? CRM cleanup? Transaction coordination? Open house systems? Hiring an assistant? Training your team? Explaining AI to clients? If something in your business feels repetitive, confusing, messy, or unnecessarily manual, I want to know about it.

The Often Letter publishes weekly and it is 100% free. Subscribe if you want practical AI workflows, agent-ready tools, and field-tested systems for real estate work. And if you have a pain point we should tackle next, reply and send it over. The best issues usually start with a real agent saying, “This part of my business is driving me crazy.”

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.

Client notes sit in your CRM. Showing feedback lives in text messages. Listing ideas hide in your camera roll. Market stats come from MLS searches, lender emails, and half-remembered conversations. Seller objections disappear into inbox threads. Open-house conversations stay in your head until the next open house pushes them out.

Then you sit down to work and rebuild the context from scratch.

You ask AI for a listing description, but the chat window has never met the seller. It does not know why they are moving, what kind of buyer should fall in love with the house, which room gets the best morning light, or what buyers said during the first weekend of showings.

You ask AI for a seller update, but it has no memory of what you promised last week.

You ask AI for content, but it has no feel for your market, your active listings, your voice, your past clients, or the questions buyers keep asking you.

So the output comes back clean, polished, and strangely empty.

That problem has less to do with AI’s writing ability and more to do with the way most agents use it. A blank chat box has no business memory. It can only work with whatever you paste in that moment.

The old real estate playbook is breaking. The next advantage will not belong only to agents who work harder. It will go to agents who learn earlier how to give AI a second brain for their business.

A second brain for real estate means your AI can remember the market with you. It can read your client notes, listing notes, open-house feedback, market themes, content ideas, and weekly reviews. It can find the pattern you forgot to write down as a strategy. It can turn scattered context into follow-up, marketing, seller updates, and better decisions.

That takes more than another prompt list. It takes a business memory AI can actually read.

Hermes and Obsidian give you a practical way to build that memory.

Obsidian gives your business a local, linkable knowledge base. Hermes gives you an AI agent that can work across that knowledge base, search it, reason through it, and turn scattered notes into actual work.

Together, they become an AI second brain for your real estate business: a living memory of clients, listings, market patterns, decisions, and follow-up.

Why most AI use does not become a second brain

Most agents use AI the same way.

1. Open ChatGPT.

2. Paste a rough prompt.

3. Ask for a listing description, caption, email, or script.

4. Rewrite half of it.

5. Close the tab.

6. Lose the context.

7. Do the same thing next week.

That workflow can save time. It rarely compounds.

Every chat starts cold. Every task makes you explain the client, the property, the market, the tone, the goal, the constraints, and the audience all over again. AI may help you move faster for one task, then forget everything that made the task specific.

A second brain works differently. It accumulates.

A good agent remembers that a seller is anxious because their last listing expired. A good agent knows that buyers in one neighborhood care more about school pickup traffic than square footage. A good agent hears the same objection five times and realizes it should become a market explainer. A good agent turns repeated questions into content, follow-up, and trust.

If that knowledge stays in your head, phone, inbox, and random chat threads, AI has no way to use it.

Hermes and Obsidian change the shape of the work. You stop asking AI to improvise from a blank screen. You give it a map of your business.

The core idea: make your business readable to AI

Obsidian is a note-taking app, but the app is the least interesting part. The real value is the folder of plain Markdown files underneath it.

Plain text is durable. It is easy to search. It is easy to move. It is easy for AI to read. Obsidian adds links between those files, so one client can connect to one open house, one listing, one market theme, and one follow-up plan.

Your CRM still matters. It should remain the system of record for contacts, pipeline stages, tasks, compliance notes, and reminders. Obsidian can become the second-brain layer around the CRM.

Use it to organize the meaning of your business:

• who the client is

• what they care about

• what happened last time

• which listings are active

• which questions keep repeating

• which objections need a better answer

• which content ideas came from real conversations

• what your market is teaching you this month

Hermes can operate on top of that vault. You can ask it to search your notes, connect related files, draft follow-up, summarize client history, turn showing feedback into a seller update, or build a weekly content plan from what actually happened in your business.

That is where an AI second brain starts to become leverage.

The vault structure I would use for a real estate second brain

Do not start by designing the perfect system. Start with a structure you can use on a busy Tuesday.

Create one Obsidian vault called:

Real Estate Second Brain

Inside it, create these folders:

00 Dashboard
01 Clients
02 Listings
03 Buyers
04 Sellers
05 Market Memory
06 Content Ideas
07 Follow-Up
08 Templates
09 Meetings and Open Houses
10 Decisions
11 Prompts
12 Weekly Reviews

That may look like a lot, but you can begin with four folders: Clients, Listings, Market Memory, and Weekly Reviews. The rest can grow as your notes grow.

This vault becomes the place where your business memory becomes usable by AI.

Build a home note for the second brain

Create a note called:

00 Dashboard/Home.md

Put this inside:

# Real Estate Second Brain

## Active Listings
- [[Listing - 123 Maple Street]]
- [[Listing - 88 Lakeview Drive]]

## Active Buyers
- [[Buyer - Sarah Chen]]
- [[Buyer - Michael Rodriguez]]

## Active Sellers
- [[Seller - Johnson Family]]

## This Week
- [[Weekly Review - 2026-06-23]]

## Market Memory
- [[Theme - Buyers asking about rates]]
- [[Theme - Sellers resisting price reductions]]
- [[Theme - Insurance costs becoming a bigger objection]]

## Content Pipeline
- [[Content Ideas]]
- [[Newsletter Drafts]]
- [[Instagram Captions]]
- [[Market Explainers]]

Think of this as the front door to the second brain. When Hermes enters the vault, this page tells it where the live parts of the business are.

Give every important client a memory page

For every active client, create a note.

Example:

01 Clients/Client - Sarah Chen.md

Template:

# Client - Sarah Chen

## Status
Buyer lead — active search

## Contact source
Open house at 123 Maple Street

## Budget
$850k–$950k

## Areas
- Westfield
- Maple Grove
- North Hills

## What they care about
- Short commute
- Strong elementary school
- Quiet street
- Updated kitchen
- Yard for dog

## Concerns
- Worried about rates
- Unsure whether to buy now or wait
- Wants to avoid overpaying

## Conversation history
### 2026-06-20 — Open house
Sarah liked the layout but asked about traffic noise and school boundaries.

### 2026-06-22 — Follow-up call
She said she wants homes within 20 minutes of work. A friend told her prices may drop later this year, so she is hesitating.

## Follow-up ideas
- Send market explainer about waiting vs buying
- Send school district boundary resources
- Send 3 listings with quiet street / commute angle

## Linked notes
- [[Theme - Buyers asking about rates]]
- [[Market Explainer - Should buyers wait?]]
- [[Open House - 123 Maple Street - 2026-06-20]]

Now you can ask Hermes:

Read Sarah Chen's note and draft a follow-up email that addresses her concern about rates and waiting. Keep it calm, practical, and specific to her commute and school priorities.

The answer will already have the raw material that makes it specific. Over time, each client note becomes a memory page your AI can return to before it writes, summarizes, or suggests the next step.

Turn each listing into a memory hub

Every listing should have its own note.

Example:

02 Listings/Listing - 123 Maple Street.md

Template:

# Listing - 123 Maple Street

## Basic facts
- Address: 123 Maple Street
- Price: $925,000
- Beds/Baths: 4 bed / 3 bath
- Square footage: 2,410
- Neighborhood: Maple Grove

## Seller goals
- Wants strong launch week
- Wants to avoid multiple price reductions
- Moving for job relocation

## Best features
- Morning light in kitchen
- Large backyard
- Renovated primary bath
- Walkable to park
- Quiet street

## Buyer profile
Likely young family or move-up buyer who wants schools, yard, and a neighborhood feel without heavy renovation.

## Showing feedback
### 2026-06-21
Buyers liked the yard and kitchen. Two visitors mentioned the upstairs bedrooms felt smaller than expected.

### 2026-06-22
One buyer asked about roof age. Another asked whether the park gets crowded on weekends.

## Objections to answer
- Bedroom size
- Roof age
- Weekend parking near park

## Content angles
- "The kind of backyard that changes how summer feels"
- "Why quiet streets matter more than ever for move-up buyers"
- "A Maple Grove home for buyers who want less renovation and more life"

## Linked notes
- [[Seller - Johnson Family]]
- [[Theme - Move-up buyers want turnkey homes]]
- [[Open House - 123 Maple Street - 2026-06-20]]

This listing page becomes more than a place to store facts. It becomes a hub that connects the seller, the buyer profile, the showing feedback, the objections, the content angles, and the future follow-up.

From this one note, Hermes can help create a listing description, an Instagram caption, a seller update, an open-house follow-up, a neighborhood explainer, a price adjustment memo, or an email to buyer agents.

One good listing note can become ten pieces of useful work.

Capture open-house memory before it disappears

Open houses produce leads, but they also produce market memory.

Most agents lose the second part. They remember a few names, enter a few contacts into a CRM, and move on. Meanwhile, the comments buyers made during the open house could have become seller feedback, social posts, buyer follow-up, and a better answer to next week’s objection.

Create a note for every open house:

09 Meetings and Open Houses/Open House - 123 Maple Street - 2026-06-20.md

Template:

# Open House - 123 Maple Street - 2026-06-20

## Listing
[[Listing - 123 Maple Street]]

## Traffic
18 groups

## Common questions
- How old is the roof?
- Are the schools assigned or lottery-based?
- Is there room to add an office?
- How busy is the park on weekends?

## Positive feedback
- Kitchen light
- Backyard
- Quiet street
- Primary bath renovation

## Negative feedback
- Bedrooms smaller than expected
- Some concern about price compared with nearby pending sale

## Buyer leads
- [[Client - Sarah Chen]]
- [[Client - Daniel Park]]

## Content ideas
- Post: "The 4 questions buyers asked at this weekend's open house"
- Seller update: summarize traffic and objection pattern
- Buyer email: explain school boundaries and park access

## Follow-up tasks
- Send Sarah commute-focused listings
- Ask seller for roof documentation
- Prepare short school boundary explainer

Then ask Hermes:

Turn this open-house note into three things:
1. A seller update email
2. A buyer follow-up message for visitors who asked about schools
3. A LinkedIn post about what buyers are asking this week

Now the content comes from your actual market instead of a generic prompt.

Use market memory as your content engine

Most agents think they need more content ideas. Usually they need a better way to notice the questions already showing up in the business.

Create a folder:

05 Market Memory

Then create notes like:

Theme - Buyers asking about rates.md
Theme - Sellers resisting price reductions.md
Theme - Insurance costs becoming a bigger objection.md
Theme - Move-up buyers want turnkey homes.md
Theme - First-time buyers confused by closing costs.md

Each theme note should collect real moments.

Example:

# Theme - Buyers asking about rates

## Pattern
More buyers are asking whether they should wait for rates to fall before making an offer.

## Real examples
- [[Client - Sarah Chen]] asked if waiting six months might save money.
- Open house visitor at [[Listing - 123 Maple Street]] said their lender pre-approval changed after rate movement.
- Buyer consultation with [[Client - Daniel Park]] turned into a discussion about monthly payment comfort.

## Useful angles
- Waiting for a lower rate can also mean more competition.
- Payment comfort matters more than trying to perfectly time the market.
- Buyers need scenarios, not predictions.

## Content ideas
- "Should you wait for rates to drop?"
- "The question buyers are asking this week"
- "Why your monthly payment matters more than the headline rate"

## Draft prompts
Ask Hermes to create a market explainer using the real examples above. Avoid pretending to predict rates. Focus on buyer decision-making.

After a month, those theme notes become a map of what your market cares about. You can ask Hermes:

What are the top five buyer concerns showing up in my notes this month? Turn each into a newsletter idea and a short social post.

That is how daily friction becomes market authority.

What Hermes does inside the second brain

Obsidian organizes the memory. Hermes works with it.

Hermes is an AI agent you can run from your computer or messaging apps like Telegram. It can use tools: read files, search folders, write drafts, summarize documents, generate checklists, manage scheduled jobs, and work across your local knowledge base.

For a real estate agent, that means you can message your AI:

Review my notes from this week. What should I follow up on before Friday?

or:

Look at my active listings and open-house notes. Draft seller updates for each seller.

or:

Find repeated buyer questions from the past month and turn them into 10 content ideas for The Monday Market Note.

Hermes can work from the files instead of waiting for you to paste everything into one chat window. Your Obsidian vault becomes the second-brain layer around your business memory.

The prompt library to start with

Use these once your vault has a few notes.

Seller update prompt

Read the listing note and recent showing/open-house notes for [listing]. Draft a seller update that explains what happened, what buyers responded to, what objections appeared, and what I recommend next. Keep it clear, calm, and specific.

Buyer follow-up prompt

Read the client note for [buyer]. Draft a follow-up message that references their priorities, answers their main concern, and suggests the next best action. Give me both text-message and email versions.

Market content prompt

Search my market theme notes and recent client conversations. Identify three questions buyers or sellers are asking repeatedly. Turn each into a short LinkedIn post and a newsletter section.

Listing marketing prompt

Read the listing note for [address]. Create 5 marketing angles, 3 Instagram captions, 1 LinkedIn post, and 1 email to buyer agents. Use the buyer profile and showing feedback from the note.

Weekly review prompt

Review all notes created or edited this week. Summarize the most important client updates, follow-up risks, content opportunities, and decisions I need to make next week.

Past-client reactivation prompt

Using my market themes and recent notes, draft a useful past-client email that gives them one timely market insight and invites a low-pressure reply.

Start small and let the second brain compound

Hermes and Obsidian are useful because they make your business legible to AI. The agents who win the next era will build AI into the way their business remembers, thinks, follows up, markets, and learns.

That advantage will feel optional right before it becomes normal.

Start with one listing note. Add one client memory page. Run one weekly review. Then let the second brain grow from the work you already do.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.

The agents who get value from AI first usually focus on one repeated part of the business: a listing launch, a seller update, a showing follow-up, a market explainer. Then they turn that work into a cleaner workflow.

That matters because AI is already moving from curiosity to baseline. NAR’s 2025 Technology Survey found that 46% of agents who are REALTORS® report using AI-generated content, including listing descriptions. Twenty percent use AI tools daily. Another 22% use them weekly. At the same time, 32% have not yet used AI in their business.

That gap is the opportunity. AI creates leverage when the agent can write faster, follow up sooner, summarize better, and stay visible more consistently. That agent gets more chances to be useful.

Who this guide is for

You are a real estate agent, team lead, broker, marketer, or transaction-side operator. You already have the normal real estate workload: listings to prepare, sellers to update, buyers to follow up with, leads to sort, showings to recap, market shifts to explain, documents to review, social content to keep alive, and clients who expect speed and judgment at the same time.

You may have tried ChatGPT once or twice. Maybe it gave you generic listing copy. Maybe it sounded too polished. Maybe it created something useful, but you did not know how to make the result repeatable. This guide is for that moment.

If you follow the steps below, you will end up with one AI workspace you can trust, one prompt library you can reuse, one listing workflow that saves time, one follow-up workflow that sounds human, one review checklist that keeps you out of obvious trouble, and one simple weekly habit that compounds. The goal is to remove drag from the work you already do.

The mistake most agents make

Most agents start with a reasonable-sounding question: “What is the best AI tool for real estate?” The question feels practical, but it usually leads to scattered testing. One week it is ChatGPT, the next week it is a video avatar tool, then a CRM add-on, then an image generator, then a browser extension nobody checks after Friday.

The better question is simpler: “Which repeated workflow costs me time every week, and what would a better version look like?” Start with the bottleneck. Then choose the tool.

Before you begin: the rules

AI can help with drafts, summaries, options, outlines, reminders, and first-pass organization. Professional decisions still belong to the agent, the team, and the brokerage review process.

Before using AI in real client work, set these rules:

1. Avoid pasting confidential client information, financial details, private transaction documents, or sensitive personal data into an unapproved public AI tool.

2. Review every fact before anything goes to a client, MLS, website, ad, email, or social channel.

3. Check listing and marketing language for fair-housing risk.

4. Keep brokerage, MLS, advertising, licensing, and local compliance rules in the loop.

5. Use AI to prepare better work while keeping professional judgment in charge.

NAR has been clear on the direction: AI is already embedded in how many agents work, but brokerages need guardrails around accuracy, privacy, fair housing, and human oversight. That is the right frame. Move faster, but keep judgment human.

What you need before starting

A simple stack is enough: one general AI assistant such as ChatGPT, Claude, Gemini, or Copilot; one place to store prompts; one folder for examples; and one review checklist for facts, tone, compliance, and next action. Use the tool you will actually open.

Step 1: Choose one workflow

Pick one workflow from this list and leave the rest alone for now:

1. Listing launch package

2. Seller weekly update

3. Showing follow-up message

4. Open-house follow-up sequence

5. Buyer consultation prep

6. Local market explainer

7. Lead reactivation message

8. Price-reduction conversation prep

9. Social post from market notes

10. Contract-to-close reminder summary

If you are unsure, start with the listing launch package. It touches the work agents already do: property notes, MLS language, seller communication, social copy, email blurbs, open-house language, and local positioning. It also teaches the most important AI habit: give the model better raw material.

Step 2: Create your AI brief

Most bad AI output comes from weak input. Agents often type, “Write a listing description for 123 Main Street,” and the model fills in the blanks with generic charm. It may invent lifestyle claims, flatten the positioning, and sound like every other listing description on the internet.

Instead, create a reusable AI brief:

Property:

[Address or internal label]

Basic facts:

[Bedrooms, bathrooms, square footage, lot size, year built, property type]

Verified features:

[Kitchen, flooring, roof, HVAC, windows, garage, outdoor space, updates]

Location context:

[Nearby amenities, commute notes, schools only if verified and allowed by your rules]

Seller priorities:

[Speed, price, positioning, privacy, timing]

Buyer audience:

[Describe needs without protected-class assumptions]

Tone:

[Warm, premium, concise, straightforward, local, luxury, first-time buyer friendly]

Channel:

[MLS description, email, Instagram caption, flyer copy, open-house handout]

Compliance notes:

[Avoid protected-class language. Do not invent facts. Flag risky claims.]

Next action:

[Schedule showing, reply for details, attend open house, request disclosures]

Save this brief. It becomes your input layer, and the more consistent your input layer becomes, the more consistent your AI output becomes.

Step 3: Build the listing launch workflow

Here is the first workflow to test. Give your AI assistant the property brief, then ask for a launch package:

You are helping me prepare a real estate listing launch package.

Use only the facts I provide. Do not invent square footage, school information, neighborhood claims, commute times, renovation dates, HOA details, or market statistics.

Create the following:

1. MLS-style listing description, concise and polished
2. Short website description
3. Seller-facing summary explaining the positioning
4. Email blurb for my buyer list
5. Three social captions in different tones
6. Open-house talking points
7. A list of missing facts I should verify
8. A fair-housing and compliance review of the language

Property brief:

[Paste your brief]

The important parts are not the outputs themselves. The important parts are the constraints: use only provided facts, avoid invention, list missing facts, and review the copy for fair-housing and compliance risk. That turns the AI from a magic writer into a drafting assistant with guardrails.

Step 4: Add the human pass

Treat the first draft as raw material. Your human pass should answer five questions: Is every fact verified? Does the copy sound like you or your team? Does it avoid protected-class assumptions or steering language? Is the strongest property detail visible early? Does the next action feel clear?

Use this review prompt:

Review the copy below as a cautious real estate marketing editor.

Check for:

1. Invented or unsupported claims
2. Vague filler language
3. Fair-housing or protected-class risk
4. Claims that require verification
5. Overused real estate phrases
6. Missing call to action

Then rewrite it to sound more specific, more human, and more concise.

Copy:

[Paste draft]

This is where agents win. The AI gives you a draft; your judgment turns it into professional communication.

Step 5: Save what works

Most agents waste AI time because they never save their best prompts. Every time you get a useful result, save three things: the prompt, the input brief, and the final edited output.

Create a folder or document called AI Workflows That Worked, then use this structure:

Workflow name:

[Seller weekly update]

Use when:

[Every Friday or after meaningful listing activity]

Inputs needed:

[Showings, feedback, market context, recommended next step]

Prompt:

[Saved prompt]

Review checklist:

[Accuracy, tone, compliance, next action]

Best example:

[Final edited version]

This is how AI becomes an operating system instead of a novelty. You are building reusable workflows, not collecting random prompts.

The Often take

AI in real estate will not arrive as one dramatic replacement event. It will arrive as dozens of small workflow upgrades. The listing description gets drafted faster. The seller update gets clearer. The open-house follow-up becomes more specific. The local market explanation becomes easier to write. The agent spends less time staring at a blank page and more time using judgment.

That is the practical opportunity. AI should become your backstage assistant, not your brand. The front stage is still trust, timing, local knowledge, negotiation, and client care. The agents who understand that balance will move faster without sounding automated.

Try this before next week

Pick one listing, one seller, or one buyer conversation. Create the input brief, run the prompt, edit the result, and save the version that worked.

That is the first step: a new workflow before a new tool.

The 30-day version

If the first week works, expand slowly: Week 1 for listing launch packages, Week 2 for seller weekly updates, Week 3 for showing and open-house follow-up, and Week 4 for local market explainers and newsletter content.

By the end of 30 days, you should have four working AI-assisted workflows. That is enough to change the rhythm of the business because it removes the blank page from the work that repeats.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.

Can people find you on Google?

Now the question is changing:

Can AI find you, understand you, and recommend you?

That sounds like a subtle shift. It is not.

For real estate agents, property managers, mortgage brokers, and title companies, local visibility has always been about trust. Who shows up when someone searches “best listing agent in Austin,” “property manager near me,” “mortgage broker for first-time buyers,” or “title company for investors”?

For twenty years, Google owned that moment.

Now ChatGPT, Gemini, Claude, Perplexity, Grok, Copilot, and AI-powered search tools are moving into the same territory. Consumers are already using AI to understand neighborhoods, compare options, estimate costs, write questions, and shortlist service providers.

They may not say, “I found you through AI.”

But before they call you, there is a growing chance they asked an AI tool one of these questions:

• “Who are the best real estate agents in Scottsdale?”

• “What should I look for in a property manager in Dallas?”

• “Compare local mortgage brokers near me.”

• “Which title company is good for investor transactions?”

• “Is this property management company reputable?”

• “What are alternatives to this agent?”

That is AI visibility.

And it is quickly becoming the next layer of SEO.

Why this matters now

Traditional SEO is built around web pages. You publish content, optimize your site, earn links, collect reviews, and hope Google ranks you.

AI visibility works differently.

AI tools do not only return a list of blue links. They synthesize an answer. They summarize what they find. They compare businesses. They cite sources when they have web access. They may mention two or three names and leave everyone else out.

That creates a new kind of winner-take-most visibility.

On Google, being result #7 may still get you some traffic. In an AI answer, there may be no #7.

There may be three recommended companies, a short paragraph, and a few citations. If you are not in the answer, you effectively do not exist in that moment.

For local real estate businesses, this matters because trust is often formed before the first phone call. If AI tools repeatedly see your competitors in directories, reviews, local articles, comparison pages, niche service pages, and well-structured websites, those competitors have a better chance of being mentioned.

Not because they are necessarily better.

Because the machines have more to work with.

How AI visibility works

AI answer engines build responses from a few different layers.

Some tools rely mostly on their training data. Others search the live web. Some combine both. The details vary by platform, but the pattern is similar.

AI systems look for signals that help them answer questions confidently.

For real estate businesses, those signals usually include:

• Your website

• Google Business Profile

• Reviews and review velocity

• Third-party directories

• Local publications

• Chamber of commerce or association pages

• Social profiles

• Schema markup on your site

• Service-area pages

• FAQ content

• Market guides

• Comparison content

• Consistent name, address, and phone information

• Clear descriptions of who you serve and what you do

The more consistent and specific these signals are, the easier it is for AI to understand your business.

That last phrase matters: understand your business.

A lot of real estate websites are technically online but semantically thin. They say things like “Your trusted local expert” or “Full-service real estate solutions,” but they do not clearly answer the questions AI tools are being asked.

For example:

• What neighborhoods do you serve?

• Do you specialize in sellers, buyers, investors, relocation, luxury, rentals, property management, VA loans, commercial closings, investor closings?

• What makes you different from nearby competitors?

• Do you have proof?

• Are there reviews?

• Are there third-party mentions?

• Is the business information consistent across the web?

• Does your site have structured data that machines can read?

• Are your pages specific, or are they all generic?

AI visibility improves when your public presence gives answer engines enough credible material to describe, compare, and recommend you.

A field example: the property manager AI could not confidently recommend

We recently looked at a property management company in a competitive metro market. I will keep the company name private, but the pattern is useful.

The business was real. It had a website. It had service pages. It had some public presence. A human could look at it and understand the basics.

But when viewed through an AI visibility lens, the gaps became obvious.

The company’s branded visibility was much stronger than its discovery visibility.

In plain English: if someone already knew the company name and asked about it, AI tools and search systems had enough information to respond.

But if someone asked a generic discovery question like:

• “Best property management companies in [city]”

• “Who should I hire to manage a rental property in [city]?”

• “Top-rated residential property managers near [neighborhood]”

• “Compare property management companies in [city]”

The company was much less likely to surface.

That is a common problem.

Many local businesses are findable when searched by name. Far fewer are visible when the buyer has not yet chosen a name.

And discovery visibility is where new business happens.

The audit showed several familiar issues:

• The website had useful information, but not enough answer-style content.

• Service-area coverage was thinner than competitors.

• Third-party sources and local citations were not strong enough.

• Schema and entity signals could be clearer.

• Reviews existed, but they were not packaged into a machine-readable trust story.

• Competitors appeared in more listicles, directories, and local pages.

• The site did not fully match the language prospects use when asking AI for help.

None of this means the company was bad.

It means the internet had not been organized in a way that made the company easy for AI to recommend.

That is the new SEO problem.

AI visibility is not “AI tricks”

There will be plenty of bad advice in this category.

People will try to sell shortcuts: “Prompt ChatGPT to rank you,” “submit your site to AI,” “stuff your pages with AI keywords,” “create 500 neighborhood pages overnight.”

Most of that will not age well.

AI visibility is not about tricking the model.

It is about making your business easier to verify.

For agents, property managers, mortgage brokers, and title companies, the best strategy is boring in the right way:

• Be specific.

• Be consistent.

• Be useful.

• Be cited.

• Be reviewed.

• Be easy for machines and humans to understand.

The businesses that win will not necessarily be the ones publishing the most AI-generated blog posts. They will be the ones with the clearest public evidence of expertise, service area, reputation, and fit.

Best practices for real estate AI visibility

1. Build pages around real client questions

Most real estate websites are built around services. AI answers are often built around questions.

That means your site should answer things like:

• “How much does property management cost in [city]?”

• “What should a seller do before listing a home in [neighborhood]?”

• “What does a title company do in a cash investor closing?”

• “How do first-time buyers choose a mortgage broker?”

• “What are landlord responsibilities in [state]?”

• “How long does it take to close in [county]?”

• “What documents do I need before applying for a mortgage?”

These pages should not be fluff. They should be specific, local, and practical.

2. Make your service areas explicit

AI tools struggle with vague geography.

“Serving the greater metro area” is weaker than a clear list of cities, neighborhoods, counties, and property types.

For example:

• Property management for single-family rentals in Plano, Frisco, McKinney, and Allen

• Listing agent for luxury homes in Paradise Valley and Scottsdale

• Mortgage broker for self-employed buyers in Orange County

• Title services for investor purchases in Harris County

Specificity helps humans. It also helps machines.

3. Strengthen your third-party footprint

AI systems often lean on sources outside your own website.

That means your visibility is affected by:

• Google Business Profile

• Yelp

• BBB

• Realtor.com

• Zillow

• Homes.com

• Expertise-style listicles

• Local publications

• Association pages

• Chamber pages

• LinkedIn

• Facebook

• YouTube

• Industry directories

• Review platforms

You do not need to be everywhere. But the main sources should be accurate, consistent, and complete.

4. Add schema markup

Schema is structured data that helps machines understand what your business is.

At minimum, real estate businesses should consider schema such as:

• Organization

• LocalBusiness

• RealEstateAgent

• ProfessionalService

• MortgageBroker, where applicable

• Service

• FAQPage

• Review or AggregateRating, when used correctly

• areaServed

• sameAs links to official profiles

Schema will not magically make AI recommend you. But it reduces ambiguity. And ambiguity is the enemy of AI visibility.

5. Create comparison and alternative content carefully

People ask AI to compare options.

You can help by publishing fair, useful comparison content. Not hit pieces. Not “why we are the best” pages pretending to be objective.

Good examples:

• “How to compare property management companies in [city]”

• “Questions to ask before hiring a listing agent”

• “Mortgage broker vs. big bank: what buyers should know”

• “What investors should ask a title company before closing”

• “Property management fees in [city]: what is normal?”

This kind of content makes you part of the evaluation process before the prospect has chosen a provider.

6. Package reviews into proof

Reviews matter, but they should not sit in isolation.

Use them to support specific claims:

• Fast communication

• Investor experience

• Tenant placement

• First-time buyer education

• Smooth closings

• Complex title issue resolution

• Relocation support

• Luxury marketing

• Local neighborhood expertise

A page that says “Read our reviews” is fine. A page that explains what clients consistently praise, links to original review sources, and matches those themes to services is better.

7. Keep business information consistent

AI tools do not like messy identity signals.

Check that your name, address, phone number, website, service areas, categories, and social links are consistent across major profiles.

This is basic local SEO, but it matters even more when AI systems are trying to decide whether two mentions refer to the same business.

8. Publish content that deserves to be cited

AI tools cite useful pages.

Useful pages usually have one or more of these traits:

• Local data

• Clear definitions

• Step-by-step explanations

• Original checklists

• Pricing guidance

• Comparison frameworks

• Timelines

• Regulations explained in plain English

• Neighborhood-specific knowledge

• Frequently asked questions

Generic “real estate tips” will not do much. Specific, local, answer-ready content has a better chance.

The bottom line

AI visibility is not replacing SEO. It is becoming the next layer on top of it.

The old work still matters: clear websites, good reviews, strong local pages, accurate profiles, useful content, and third-party credibility.

But the output is changing.

People are no longer only scanning search results. They are asking AI tools for recommendations, comparisons, explanations, and shortcuts.

That means your online presence has a new job.

It has to persuade humans.

And it has to be understandable to machines.

If AI tools cannot clearly tell who you serve, where you work, what you do, and why you are credible, they may skip over you and recommend someone else.

Not because the other business is better.

Because the other business is easier to understand.

RECOMMENDED RESOURCE

Build your first AI team

If this article was useful, you may also like Your First AI Team — a practical guide to building AI employees with clear roles, tools, memory, and workflows.

It is written for operators, founders, and professionals who want to stop using AI as a one-off chatbot and start managing it like a coordinated team.

Read the book on Amazon →

A short practical guide for operators building with AI.