Prompts
In short: here are 88 prompts you can paste into ChatGPT, Claude, or Gemini, one for each use case in this directory, grouped into 10 categories. Each prompt comes from a real, sourced example of an agent, team, or investor using AI. Follow the link to see who did it, which tools they used, and the original post. The prompts were written by this directory, inspired by each source. They are not quotes. Replace the [bracketed] parts before you run one.
All lead follow-up & inside sales use cases →
How many homeowners in [ZIP code] have owned for 10+ years, have more than 50% equity, and are over 65? Then: Tell me about [address]. What's owed on it, and what have similar homes sold for? Pull the sale and loan history. Then: Save this search as an audience called [farm name] and watch it.
[Attach screenshots of the text thread with the new lead.] Create this lead in Follow Up Boss. Use the name and phone number shown in the screenshots, set the source to [source], and log a note that summarizes what the lead asked for, their timeline, and any property or price they mentioned. If a name or number is not visible in the screenshots, ask me instead of guessing. Show me the contact and the note before you save them.
Sphere Signals bot, first job: Here are 30 top names from my database: [paste names and towns]. Search the open web for anything about them. Give me only what you can link to, ordered by how good a reason it is to pick up the phone. Skip anything older than 90 days, and skip anything you can't source. Then: Every Monday at 7am Eastern, run that same search on the list for the past seven days and post what you find here with links. If there's nothing, post nothing. Sphere Desk bot, first job: Go back through the last 30 days of my inbox. Give me two lists: people who emailed me and never got a reply, and past clients or referral sources who've gone quiet. For the top five in each, tell me the last thing we actually talked about and draft a two-sentence reply I could send today. Nothing that sounds like a newsletter. Then: Every weekday at 7am, run that check on the last 48 hours plus anyone crossing 90 days of silence. Post it here. If there's nobody, post nothing.
Draft a viewing proposal email and a short SMS for this lead. Lead: [name], [email], [phone] Property: [address or listing URL] Proposed slots: [slot 1], [slot 2], [slot 3] Calendly booking link: [url] Agent: [name], [brokerage], [phone] Email rules: warm and specific, name the property, list the three slots as bullets, include the Calendly link once, ask them to book or reply with a preferred time, under 180 words. SMS rules: under 300 characters, name the property, include the Calendly link, end with Reply STOP to opt out. Do not invent availability outside the slots provided.
You are a listing WhatsApp assistant for [brokerage] in [market]. Approved listing facts only (from Airtable): [address, beds, baths, price, status, showing notes]. Lead message: [paste]. 1. Answer only with facts from the listing record. If the answer is not in the record, say you will have an agent confirm. 2. Ask one qualifying question (timeline, budget band, or financing status). 3. Classify the lead as hot, warm, or cold with a one-line reason. 4. If hot and they want a viewing, propose times only from this availability list: [calendar slots] and ask them to confirm one. Never invent amenities, prices, or open times.
From the attached list of properties in [farm area]: 1. Keep only properties last purchased [7] or more years ago. 2. For each remaining property, look up the owner on the [county] property appraiser website and record the owner mailing address. 3. Add a column "Absentee owner" = Yes when the mailing address differs from the property address. 4. Output a spreadsheet with address, purchase year, owner mailing address, and absentee flag. Note any record you could not verify instead of guessing.
Search my messages from the last 12 months. Find every conversation where the other person mentioned a life event: a baby, an engagement or wedding, a new job, a move, a renovation, a kid starting school, a death in the family. For each one give me their name, the date, one line on what they said, and how long it's been since we last talked. Sort by how long it's been. Don't draft anything. Skip group threads.
You are the operations assistant for [brokerage name]. For each agent below, review the agent's assigned CRM leads (attached) and produce a morning briefing: 1. The 5 highest-priority leads, ranked using these factors: buyer or seller, lead source, geographic focus, price point, response timing, and last call duration. 2. For each lead, one suggested follow-up approach based only on the CRM notes. 3. Any lead with no contact in [X] days. Keep each briefing under 200 words and address each briefing to the agent by first name. Agents and leads: [paste]
Pull my Granola notes from today's call with [owner name] about [property address], plus my earlier notes and CRM history for [owner name]. 1. Summarize the owner's situation: intent to sell, view of value, debt maturity, recent capital expenditure, and market concerns. Mark each item as stated on the call or inferred. 2. List every objection and every commitment either of us made, such as comps I promised to send. 3. Note anything that changed since our earlier conversations. 4. Recommend the single clearest next step. 5. Draft a follow-up email of three short paragraphs that delivers what I promised and proposes the next step. Use only facts supported by the call or the earlier notes, and leave out personal or family details. Save the email as a Gmail draft. Do not send the email.
Using the contact files in [folder]: 1. Find 10 past clients or sphere contacts I have not spoken to in 6 months or more. For each contact, draft a short, warm note in my voice that references our last real conversation. 2. Give me 5 people to call this week, with two lines of prep on why now. 3. Flag birthdays and home closing anniversaries in the next 14 days and draft a one-line note for each. Put everything in one document named "Monday Sphere Brief [date]". Draft only. Send nothing.
A homeowner submitted this seller form: [paste address, beds, baths, square feet, condition, timeline, reason for selling]. Here are verified comparable sales from the MLS: [paste 3-6 comps with address, sale date, price, beds, baths, square feet]. 1. Qualify the seller lead: timeline, motivation, and next step, in 3 bullets. 2. Write a one-page "CMA-lite" for the homeowner: a suggested price range based only on the comps above, how each comp compares to the subject home, and a clear note that a full CMA after a walkthrough will refine the range. Do not invent comps or market statistics.
Run my planning-filings lead skill on [city planning portal URL]. 1. Collect the [5] newest case filings. For each case, record the address, applicant, representative, case number, filing date, project description, and requested entitlement. 2. Open the initial submittal documents and project plans, and record any owner, architect, engineer, surveyor, or general contractor named there, with the contact details the documents show. 3. Append each case as a new row in my existing Google Sheet [sheet name]. Do not create a new sheet. 4. Add the source link for every row and a research note on anything you could not confirm. 5. If a document won't load, fall back to values you already captured and fill as many fields as possible instead of leaving the row blank. Use only contact details published in the filing documents.
Parse this property-portal lead notification email into structured fields. Email body: [paste]. Return JSON with: lead_name, lead_email, lead_phone, property_interest, budget, timeline, portal_source, lead_score (Hot|Warm|Cold), score_reasoning (1 sentence), acknowledgment_email (under 120 words confirming we received the inquiry and naming the property if present). Scoring rules for my market: Hot = ready to tour within 2 weeks or pre-approved; Warm = 30–90 day timeline; Cold = researching or no timeline. Do not invent contact fields that are missing from the email.
You score real estate inquiries. Return only JSON with these keys: intent (buyer/seller/both), summary (one sentence), timeline_band (0-3 months / 3-6 months / 6+ months / unknown), financing_readiness (pre-approved / cash / not started / unknown), score (1-10), recommended_next_step. Scoring rubric: [your rubric, e.g. +3 for timeline under 90 days, +2 for pre-approval, +2 for a specific area and budget]. Base every field only on the inquiry text below. Never infer a score from the person's name, age, or other personal characteristics. Inquiry: [paste form submission]
You are my open house follow-up assistant. Listing: [address], listing URL: [URL], open house date: [date]. 1. Read every name, email, and handwritten note on the attached sign-in sheet photo. List any entry you cannot read instead of guessing. 2. From the listing URL, pull 4 standout property features. 3. For each visitor, draft a 90-110 word email that thanks the visitor, references one note I wrote about that visitor, names two property features that match the note, asks one open question, and offers a second showing or market alerts. 4. Use a warm, plain tone, vary the opening line, and sign as [my name]. Show all drafts and wait for my approval. Do not send any email.
Write the first text message for a new buyer inquiry. Lead data: name [first name], property [address], list price [price], MLS # [number], source [Zillow/IDX/Facebook]. Rules: under 300 characters, mention the exact property address, ask one question about the buyer's timeline or a showing time, include my name [name] and brokerage [brokerage], end with "Reply STOP to opt out." Then write 4 follow-up questions an AI qualification sequence should ask over the next 3 days (timeline, budget range, pre-approval status, showing availability), one per message.
Here are yesterday's expired and withdrawn listings in [zip codes]: [paste address, list price, days on market, price changes, listing remarks]. For each listing: 1. Give a motivation score 1-5 based only on days on market, number of price reductions, and status. 2. Write a 120-word first email to the owner that names the property address, gives one specific reason the listing may not have sold (based on the data provided), and offers a no-pressure pricing review. 3. Write a 2-email follow-up sequence (day 3 and day 7). Do not claim a sale price, buyer, or result you cannot support.
Here is an export of my CRM leads with last activity, source, stage, and notes: [paste CSV]. 1. Score each lead 1-10 for likelihood to transact in 90 days, with a one-line reason drawn only from the notes and activity. 2. For the top 10 leads, draft a 2-3 sentence follow-up text or email in my voice that references the last real conversation in the notes. 3. Create today's task list: lead name, channel (call/text/email), and the reason to reach out today. Draft only; I will review and send.
Read these two listing presentations. Don't summarize them. Write me a reusable skill. Start with a name and a one-line description of when it should be used, then the instructions: every section in order, what belongs in each one, roughly how long each one runs, and the rules my writing follows, including sentence length, how formal it is, the words I never use, and the ones I lean on. Be specific enough that someone else could build a presentation in my style from this document alone.
Save this as a skill named "MLS description". Inputs I will give you: property address, executed listing agreement (PDF), and MLS tax roll record. Steps: 1. Extract beds, baths, square footage, lot size, year built, and legal details from the tax roll; flag any conflict with the listing agreement. 2. Write an MLS public remarks paragraph within [character limit] characters using only verified facts, plus private agent remarks with showing instructions from the listing agreement. 3. List every fact you used and the source document for each fact. Never add features that are not in the documents.
Go to the official town website for [town, state] using the browser in the Claude desktop app. Find the Planning Board, the Zoning Board of Appeals and the Conservation Commission. Read every agenda and set of meeting minutes posted in the last twelve months. List any application, hearing or decision involving a property within about half a mile of [subject address]. For each one give me the address or parcel ID, the date, what was proposed, and where it stands now. If a document won't open, or you can't find a board's page at all, tell me that instead of guessing. Then: Of everything you found, which items would a buyer's attorney raise at closing, and which would change what someone is willing to pay for this house? Rank them and give me one sentence on each.
Read the search results on this page. Build a table with address, status change, list price, price per square foot, and days on market. Then write me four sentences in plain language: what moved this week, which price cut matters most and why, which property has sat long enough that I should call the agent, and one specific thing I could say to a seller about this week's activity. No preamble, no summary at the end.
You are my listing strategist. Analyze [address], [beds/baths/sqft]. The sellers want to list at [$their number]. My three closest comps are [comp 1: price, size, condition, distance], [comp 2], [comp 3]. Give me the price you would recommend, the two strongest data points behind it, the exact objection these sellers will raise, and word for word how I answer it. Be specific to these comps, not generic.
Build a single-page website for my listing at [address]. Use the attached reference file for layout and style, and use only the attached photos and the facts below: [price, beds, baths, square footage, key features]. Include my name, my brokerage name, my license number, and a schedule-a-tour form with fields for name, phone, email, and preferred times. Make the page work on mobile and desktop. Do not invent any fact about the home, the neighborhood, or the schools; leave a bracketed blank where a fact is missing. When the page is done, tell me what I need to connect the form to [Tally or my form tool] so submissions email me and reach my CRM.
I am attaching photos of each room plus these property facts: [address, beds, baths, square feet, upgrades]. 1. For each photo, list the visible features (finishes, fixtures, light, layout). Do not guess at features that are not visible. 2. Write a listing title (under 60 characters), a 200-word description, and 5 key selling points. 3. Using only these verified comps [paste], suggest a list price range with a one-line reason per comp. Mark every claim that needs agent verification.
Build a single-page interactive broker opinion of value for [property address, unit count]. Inputs with sliders: current gross rents [$], vacancy [%], operating expenses [$], annual rent growth [%], exit cap rate [%]. Outputs that update live: NOI, value at cap rate, price per unit, and a 5-year value table. Use plain HTML, CSS, and JavaScript in one file, my brand colors [hex], and a disclaimer footer: "Broker opinion of value, not an appraisal."
Turn this CMA data into a branded one-page seller report: [paste subject property and 4-6 comps]. Brand: [brokerage name], colors [hex codes], font [font], logo [attached]. Sections: headline value range, comps table (address, sale date, price, price per square foot, days on market), 3 plain-English takeaways, and a recommended list price with reasoning tied to specific comps. Format as a print-ready one-page HTML document. Use only the numbers provided.
Using my template [file] and the CMA export [file] for [address]: Build a listing presentation PDF with these sections in the template layout: my bio, the comps table, a neighborhood market snapshot from the CMA data, a pricing strategy with three price scenarios, and my marketing plan [paste plan]. Use only numbers from the CMA file. Save as "[address] Listing Presentation.pdf" and list any section where data was missing.
I am virtually staging a listing photo. Room: [living room/bedroom], dimensions: [size], style of the home: [e.g. 1960s ranch with oak floors], price point: [range]. Write a staging brief for the AI tool: furniture pieces, scale, color palette, and the elements that must stay unchanged (windows, floors, walls, fixtures, views). Then write the photo caption disclosure: "Virtually staged. Furniture is not included."
Write a listing description for the property below. Use only these facts: [beds, baths, square feet, lot, year built, upgrades, amenities, building features, nearby transit and landmarks].
- 180-220 words, lead with the strongest feature.
- Include these search phrases naturally: [e.g. "two-bedroom condo in [neighborhood]", "doorman building"].
- No clichés ("nestled", "boasts", "dream home"), no exclamation marks.
- Describe the property, not the ideal buyer.
Then give a 40-word version for social media. All marketing & content use cases →
Look at my last 30 days of Instagram posts. Which three got the most engagement, and what do they have in common? Be specific about the format, the topic and how each one opened. Here is next week: Listing: [address], [price], open house Sunday 1 to 3. Closing Thursday, buyers' first home. Noticed: [one thing about your market]. Using what worked, plan five posts for next week. For each one, give me the day, the format, a one-line idea, and which of my facts it uses. Don't write the posts yet, and don't use any facts I haven't given you. Then: Write posts one and two in my voice. Keep each under 80 words and use only the facts I gave you.
Write voiceover scripts for these home tour videos: [paste for each home: city, price, beds, baths, 3 standout features, video length]. Each script: 25-30 seconds when read aloud (about 70 words), open with the price and city, name the 3 features in the order the features appear on screen, and end with "[call to action, e.g. DM me 'TOUR' for details]". Plain spoken language, no emojis, describe the home only.
Write a 1,200-word web page answering: "[buyer question, e.g. Which [city] builders are offering rate buydowns this month?]". Rules: answer the question in the first 2 sentences, use question-style subheadings, cite a source and date for every incentive or price, add a 5-question FAQ section, and include an author box with my name, license number [#], and brokerage. Mark every fact I must verify before publishing.
I'm a real estate agent in [city] who works [neighborhoods or ZIP codes]. I'm testing ChatGPT ads for seller leads with a budget of $[amount] a day for [number] weeks. 1) List 15 questions a homeowner in my area might ask an AI assistant before selling (home value, timing, the selling process, market conditions) that I can use as prompt targeting. 2) Write 5 one-line ad texts under 60 characters, each tied to one of those questions, with no promises about price or results. 3) Write a one-page plan for the landing page and the 6-month follow-up sequence for leads who plan to sell in 3 to 6 months. 4) Give me a weekly tracking sheet layout with spend, clicks, click-through rate, cost per click, leads, and cost per lead, with a column to compare against my Facebook and Google Ads numbers.
Listing video edit checklist for [property address]: 1. Source doc: listing description split into one short paragraph per feature (kitchen, yard, neighborhood). 2. Shape: Landscape for the website and email, Portrait for Instagram. 3. Narration: in the script editor, replace every 'stunning', 'must-see', or 'won't last' with a detail only I know, such as which way the kitchen window faces at sunset. 4. Visuals: swap every stock image for a real listing photo, and delete any scene that shows a pool, view, or feature the home doesn't have. 5. Facts: remove any school, HOA, or park claim that isn't in my source document. 6. Review: send the video to one person who hasn't seen the listing and ask whether the video looks AI-made.
Search these sources for [city] from the last 7 days only: [council agenda URL], [planning/zoning URL], [permit portal], [local news sites]. List up to 10 items about new developments, zoning changes, road projects, business openings, or housing data. For each item give: headline, date, source link, and one sentence on why a local resident would care. Exclude items older than 7 days and items without a source link. I will choose which items become posts.
The attached CSV holds my LinkedIn post history with engagement data. I want a deep performance analysis now and, after that, a reusable LinkedIn writing skill built from the findings. Ask me your clarifying questions before starting (for example, how to treat reposts and whether to rank by raw engagement or engagement per follower). Then report: my top-performing hooks, performance by post format, which posts drew the most comments versus likes, my signature phrases and formatting patterns, my posting cadence, and what the data can't tell us. Then build a skill that saves my top 20 posts as reference examples, returns two or three post options per run with a scorecard for each, and triggers on "turn this into a LinkedIn post."
Look through my last 50 buyer and client emails and list the questions people ask most about specific neighborhoods, in their own words. Then: Use OpenArt to build me a video. I sell homes in [neighborhood, city], and the question I want this video to answer is: [paste the buyer question]. Write me a 24-second lifestyle video, three shots of 8 seconds each, that answers it by showing the FEEL of living here to a [buyer type, e.g. young family / downsizing couple]. No shots of any specific house. For each shot, write one clean generation prompt: camera move, time of day, mood, and what is in frame. Keep it calm and cinematic, not flashy. Then: Using OpenArt, generate the first shot from my list as an 8-second vertical clip.
Role: You are a listing copywriter for a [city] real estate agent. Context: This copy is for [MLS remarks / an Instagram caption / a follow-up email] read by [audience]. Data: [address, price, beds/baths, square feet, standout features, one recent comparable sale]. Voice: direct, warm, specific. Never use "nestled", "boasts", or "dream home". Task: Write one [150-word MLS description]. End with one line that encourages a showing without hype.
I am building a local education page for buyers in [county/city]. Topic: "[e.g. What are closing costs in [county]?]". 1. Write the page in plain English with a worked example using these verified fee figures: [paste transfer tax rates, recordation fees, typical title fees with sources]. 2. Build a simple closing cost calculator in one HTML/JavaScript block with inputs for price, loan amount, and buyer/seller side. 3. List the assumptions the calculator makes and the sources I should link.
Using only the sources in this notebook, write a buyer guide for [area]. Sections: local landmarks and parks, popular restaurants and coffee shops, commute times to [employment centers], current price ranges by home type (cite the source and date), and a homebuying checklist for this market. Cite a source for every number. Do not rank or describe areas by who lives there. Then create a 10-question FAQ from the guide for a lead magnet.
Write a 90-second script for my AI avatar video about [topic, e.g. "5 mistakes first-time buyers make in [city]"]. Inputs: verified local data [paste stats with source and date], my talking style [short sentences, casual, one joke max]. Structure: hook in the first 5 seconds, 3-5 points with one data point each, call to action [subscribe / book a call]. Add on-screen text suggestions for each point. Do not include any statistic I did not provide.
Using the attached CSV of closed sales in [neighborhood] for the last 12 months: Build an interactive HTML dashboard with: monthly median price and price per square foot, median days on market by month, a table by building or street, the 10 fastest sales, the 10 largest list-to-sale discounts, and a scatter plot of price per square foot vs. days on market. Add a "Data source and date range" note at the top. Do not fill gaps with estimates; show missing data as blank.
All transaction coordination use cases →
Ask Gmail Live (or type into Gmail's search with AI): "For the deal at [address], what are the inspection, appraisal, financing, and closing dates mentioned in my email, and which emails are they in?" Then open each source email card and compare each date with the executed contract and any amendments before you put it on your calendar or tell your client.
Read the accepted contract, every counteroffer, and every addendum attached for [address]. Return: 1. Parties (buyers, sellers, agents, lender, title/escrow) with any contact details listed. 2. Every date and deadline, each with the document name and page or section the date came from. 3. Final agreed price, earnest money, and concessions after all counteroffers. 4. Conflicts between documents (for example two closing dates) and blank fields or referenced documents that are missing. Do not resolve conflicts yourself; list the conflicts for my review.
I uploaded this executed purchase contract: [attach PDF]. 1. Extract every contingency and hard deadline with the calendar date and which party owns the action. 2. Build a day-by-day checklist from contract to close. 3. Draft three short chase emails, each addressed to the correct party for a specific task (for example open escrow with title, request inspection report from the buyer's agent, or confirm loan commitment with the lender). 4. Flag any missing signatures, blank date fields, or conflicts you see in the document. Do not invent dates that are not in the contract. Mark every extracted date as unverified until I compare it to the PDF.
Find every executed contract, amendment, and addendum in my inbox for [property address]. Read each document and build a deadline table: deadline name, date and time, the contract section that sets the deadline, and the source document. Include option/inspection period end, earnest money due, financing approval deadline, appraisal deadline, and closing. If two documents give different dates, stop and ask me which date controls. After I approve the table, add each deadline to my calendar with a 24-hour reminder.
Transaction: [address], buyer [name], seller [name], lender [name/email], title [company/email], closing [date]. Current milestone: [e.g. day 3 after acceptance / inspection resolved / 5 days before closing]. Draft the [lender check-in / buyer inspection update / title pre-closing summary] email: - under 150 words, subject line included - list what is complete, what is due next with the date, and what I need from the recipient - no legal interpretation of contract terms. I will review before sending.
All client communication use cases →
Build me an interactive monthly payment calculator for a home buyer with sliders for price, down payment percent, interest rate and loan length. Show principal and interest, property tax, insurance, HOA dues and mortgage insurance as separate lines, then the total. Start with price [price], down payment [percent], rate [rate], property tax [rate], insurance [per year], HOA [per month, or 0]. Don't look up a rate. Use mine. Then: Build me a one-screen dashboard from this calculator that I can share with my client. Include the monthly payment broken into principal and interest, taxes, insurance and HOA; the same payment at a rate half a point higher and half a point lower; cash needed at closing with [percent] down and closing costs of about [percent]; and what $25,000 more or less in price does to the monthly payment. Label every number as an estimate, and put my name and phone number, [name, phone], at the top. Then: Add two sentences at the top, in plain English, that tell this buyer what the dashboard means for their search. Then move the number that would surprise a first-time buyer most to the top.
Using my MLS data, pull every listing change event in [subdivision] for homes between [min] and [max] square feet over the last 30 days. 1. Count the distinct listings with a change and the total change events. 2. Break the events down by type: price decreases, new listings, active under contract, back on market, on hold, withdrawn, pending, expired, closed. 3. List the three largest price cuts with the dollar amount and what happened to each listing afterward. 4. Note where my seller's listing at [address] sits: days on market and any price changes. Write the result as a short weekly email to my seller, [first name], in plain language. Report only what the MLS data shows; do not predict prices or tell the seller what to do. I will add my recommendation myself.
Build a one-page seller net sheet calculator in HTML and JavaScript. Inputs: sale price, payoff balance, [compensation figures as agreed in the listing agreement], title/escrow fees [defaults], transfer tax rate [%], prorated property taxes, HOA fees, seller concessions, and monthly carrying cost. Show two scenarios side by side (for example price A sold in month 1 vs. price B sold in month 3) including carrying costs, with estimated net proceeds for each. Footer: "Estimate only. Final figures come from the settlement statement."
I'm signed in to my MLS in this browser. My buyer wants to tour these homes: [addresses or MLS numbers]. For each property, open the listing and return a table with the address, MLS number, list price, status, the showing instructions exactly as written, how the listing agent wants showings booked (ShowingTime, text, call, or other), and the listing agent's contact details. If any field is missing, write "not found". Do not book anything or contact anyone; I will review the table first.
I'm building a weekly capital markets routine for [asset class] clients in [primary market]. Ask me clarifying questions first, then write the system prompt for the routine. The routine should build an 8 to 10 slide PowerPoint deck every Monday covering: 1. The 10-year Treasury yield and recent movement, SOFR, and the Fed target range, with notes on the last and next Fed meetings. 2. CPI and PCE. 3. Fannie Mae and Freddie Mac indicative rates for 5, 7, and 10-year fixed terms, plus bridge and balance sheet options. 4. National [asset class] cap rates by class, transaction volume, and pricing trends. 5. [Primary market] cap rates, recent comp transactions, vacancy, rent growth, and supply. Present the data without editorial opinion. Cite a dated source for every figure, and mark any figure you can't verify instead of estimating it. Save the deck for my review; do not send it.
Read this week's emails, call notes, and CRM activity for the listing at [address] [paste or connect]. Draft the seller portal update with these sections: offers received (price, terms, status), buyer activity (tours, data room views, inquiries), marketing metrics, next milestones with dates, and broker recommendation. Mark every figure with the figure's source. Do not publish; show me the draft for review.
You are an SMS listing assistant for [brokerage]. Lead mobile: [number] Known property codes for this lead: [codes] Property rows found: [paste sheet rows with code, address, beds, baths, price, status]. Inbound SMS: [paste] 1. Answer using only the property rows. If a code is missing, say which codes you found and ask them to resend the correct code. 2. If they ask to tour, check these open slots only: [calendar availability] and propose one time. 3. Keep the SMS under 320 characters when possible; split into two short messages if needed. 4. End with how to opt out (Reply STOP). Do not invent inventory, prices, or open appointment times.
Save as "Buyer: [first name + initial]". Criteria: [beds, baths, must-have features, price range, areas, deal breakers]. Status: [touring/relocating date]. When I ask "what's new for [name]", query the MLS for new active listings from the past 7 days, rank the listings by fit to the saved criteria, and explain the top 3 matches in two sentences each. Weekly: "Pull what's new for all my active buyers. Top 3 per buyer. Flag any property where two of my buyers might compete."
Review the attached home inspection report for [address]. 1. List every [electrical and plumbing] issue: report page, the inspector's wording, location, and severity (safety / major system / minor). 2. Add a rough cost range for each item and label every range "estimate, verify with a licensed contractor". 3. Summarize the total range for my buyers in plain English, under 200 words. After contractors confirm pricing [paste confirmed numbers], format the agreed repairs or credit as a draft rider to the purchase contract for attorney review.
All investing & underwriting use cases →
Audit this underwriting workbook against the offering memorandum and the stated investment terms I attached. Use only the figures in these documents. 1. Check for mechanical formula errors and broken links between tabs. 2. Find any expense that appears in the model but is missing from the operating expense total, or that is counted twice. 3. Check whether first-year revenue uses in-place rents from the rent roll or market rents, and flag any gap. 4. Compare the exit cap rate with the going-in cap rate and flag aggressive exit or growth assumptions. 5. Compare the offering memorandum, rent roll, and T-12 for conflicting unit occupancy, rents, or expenses, and flag any expense that looks unusually low for the property's revenue, such as real estate taxes. For each issue, give the sheet and cell, the dollar effect where you can calculate the dollar effect, and the source document. Do not fix the model or assume missing figures. End with a list of questions I should send the sponsor or broker.
Act as a real estate acquisitions analyst. The folder contains [offering memorandum], my underwriting model, and my buy box. 1. Enter the relevant assumptions into the blue input cells of my model only; do not edit formulas. 2. List every assumption you entered with the OM page the assumption came from. 3. Flag missing or unclear data. 4. Screen the deal against my buy box [paste criteria] and give pass / review / fail with reasons.
I'm preparing a knowledge base for an AI deal room chatbot that prospective investors will question before a call with my team. Using only the offering documents I attached, write a question-and-answer knowledge base for this deal grouped under: property, market, business plan, underwriting assumptions, sensitivity to changed assumptions, tax treatment, fees and distributions, and risks. For each answer, cite the document and page. Where the documents don't answer a likely investor question, list the question under 'Needs sponsor answer' instead of answering it. Flag any statement that could read as a promise of returns so my securities counsel can review the wording.
Read the deal folders for these active multifamily opportunities in Google Drive: [deal names]. Using our buy box file and the fund's exit strategy and target returns, stress-test the deals side by side. For each deal, list the key assumptions, the biggest risk, and how the deal ranks against the others. For any deal that fails the buy box, record the main rejection reason (location, pricing, unit mix, capital expenditure, returns, or other) in the rejection log with one sentence of evidence from the deal files. Do not change any underwriting model.
I want to build a deterministic lease-by-lease cash flow calculator. Here are 3 historical deals with inputs and the correct annual cash flow outputs: [attach]. 1. Write the calculation spec: base rent, escalations, free rent, recoveries with base-year stops, vacancy, and reversion. 2. Implement the spec in [Python/Excel formulas] and write a test that compares the outputs to my historical deals within $1. 3. Report every mismatch and the likely cause. Do not change my historical numbers to make tests pass.
Build a password-protected web app that serves as the quarterly LP update for [deal name], using only the attached investment summary, T-12, rent roll, and Excel files. Sections: 1. Summary: NOI versus underwriting variance and a health badge. 2. Performance versus underwriting: each metric shown as underwritten, actual, variance, and a short explanation. 3. Refinance and sale scenario lab with sliders for rate, exit cap, and timing; an LP enters an ownership percentage and sees their share of net proceeds. 4. Unit-by-unit stacking plan filterable by rented, month-to-month, notice to vacate, and vacant. 5. Top three risks ranked by severity and likelihood. 6. Capital plan tracker with spent, remaining, and budget variance per project. Before writing code, ask me every question you can't answer from the files. If two documents disagree on a figure such as NOI or DSCR, stop and ask which figure to use.
I want a weekly routine that tracks the comparable properties in the attached CSV against my subject property, [subject address, unit count, unit mix]. Before you build anything, ask me your questions and draft the routine prompt for my approval. 1. Review the CSV first. Flag any comp that looks like affordable, supportive, or other non-market-rate housing, and any comp whose unit types don't match my subject property. 2. For each remaining comp, track by unit type: concessions, asking rents, availability, vacancy, renovation activity, permit filings from [city permit portal], and ownership or for-sale listings. 3. Prioritize listing sites, property management sites, the city permit and planning portals, and Google News. 4. Output a change-only digest at the top (what moved since last week), then the full table with a source link for every figure. 5. Mark any figure you can't confirm, such as a listing that has been live for months, as 'verify' instead of logging that figure as fact.
This folder holds the [year] operating-expense statement and the tenant CAM lease abstracts for [property name]. 1. Review every file and summarize what you found. Before calculating anything, list each decision you can't make from the documents alone, such as missing pro rata shares, amounts billed, fee treatment, caps, and exclusions, and wait for my answers. 2. After I answer, build an Excel workbook with formulas and four tabs: Summary (each tenant's pro rata share, CAM owed, CAM billed, and balance due or refund), CAM Pool (how expenses flow into the recoverable pool, including any gross-up), Tenant Detail (the calculation for each lease), and Assumptions and Flags. 3. Run mathematical checks and show the results. If a core document is missing, stop and tell me which document instead of inventing a value.
Act as the asset manager for [property name]. I need a full [year] operating budget in Excel built from the attached deal folder. Before building, draft the prompt and ask me your clarifying questions. Use these rules unless I change them: 1. Build every line bottom-up from the source documents, and use actuals wherever they exist. 2. Known [year] figures override generic escalation assumptions. 3. Estimate lease renewals and rollovers from the market comps. 4. If a figure conflicts with the ownership directive of [directive], adjust to hit the target, flag the change, and keep the bottom-up figure visible next to the adjusted figure. Output four tabs: [year] Budget (prior budget, reforecast, new figure, variance, and a notes column citing the basis for each line), Capex Plan, Hard Actuals and Flags, and Directive Reconciliation. Confirm every file is present before you start, and label any industry benchmark you use as a benchmark.
Every Monday at 7am, search my email, CRM notes, and call notes from the last 7 days (first run: last 12 months) for rent comps. A rent comp is any mention of a specific property or unit with an asking or achieved rent. Append each comp to [sheet] with: property address, unit type, beds/baths, square feet, rent, concessions, date observed, and a link to the source email or note. Skip duplicates by address + unit type + month. Never estimate a missing rent.
We are filling my underwriting model [file] for the deal in this folder (OM, rent roll, T12). Go one step at a time and wait for my OK after each step. 1. Cell [C11]: find the property address in the offering memorandum (usually the first 4 pages). 2. Unit mix tab: list each unit layout in the rent roll under Unit Type, then fill Unit Count, Average Square Footage, and Average Rent column by column. 3. T12: map each income and expense line to my categories: [e.g. Late Fees -> Other Income; Snow Removal -> Contract Services; Office Supplies -> General/Admin]. Stop and ask when a line is unclear. When we finish, save this workflow as a Skill named "Underwrite deal".
I am evaluating a short-term rental purchase as a [buyer / LP / lender]. Property: [address], price [$], loan terms [rate, LTV, amortization]. Revenue inputs I verified: monthly ADR and occupancy for each of the 12 months [paste, with data source]. Expenses: cleaning, platform fees, management, utilities, insurance, property tax, supplies, maintenance reserve, furnishing budget. Compute monthly NOI, minimum and average DSCR, cash-on-cash, and levered IRR over a [5]-year hold. Show the 3 weakest months and a downside case with occupancy 15% lower. Do not supply any revenue or expense figure I did not give you; list the missing inputs instead.
All STR operations use cases →
Here is a guest review of my rental: [paste review]. Here is what actually happened, with dates: [paste facts, messages, and the steps we took to fix any issue]. Write a public response under 90 words that thanks the guest, states the facts calmly without arguing, mentions the fix or policy that addresses the concern, and reassures future guests. No sarcasm, no guest names, no private details.
Attached is last month's performance for my rentals: revenue, occupancy, ADR, booking lead time, length of stay, and review scores by property [paste or upload]. Compare against the same month last year [paste] and against the prior month. Give me exactly 3 revenue moves for next month, each with: the property, the evidence in the numbers, the specific change (price floor, minimum stay, gap-night rule, listing photo, amenity), and how I will measure the change. Use only the numbers provided.
When a reservation is created or changed, create or update a turnover task in [tool] for [unit]. Task fields: unit name, checkout date/time, next check-in date/time, hours available for the clean, guest count for linen count ([sets per guest]), special requests or damage notes from the last guest, and restock items flagged by the cleaner last time. If the reservation is cancelled, mark the task cancelled. If the window is under [3] hours, tag the task RUSH and notify me. Do not message the cleaner.
You answer guest messages for [property name]. Use only the facts below. Property FAQ: [check-in/out times, door code process, Wi-Fi, parking, hot tub rules, trash day, appliance instructions, local recommendations]. House rules: [paste]. Rules: reply in under 80 words, friendly and clear. Never offer refunds, discounts, early check-in, late checkout, or extra guests; instead say "Let me check with the host and get right back to you" and tag the message HOST. Tag any safety, damage, or lockout message URGENT. Guest message: [paste]
You are our brokerage marketing fact-checker and blog optimizer. Brand voice rules: [paste 5 sample paragraphs or your style guide]. Topic and outline: [paste outline]. 1. Fact-check every claim in the outline. Flag anything that needs a source link, local MLS verification, or legal/compliance review. Do not invent statistics. 2. Expand the outline into a full draft that matches our brand voice. 3. Apply SEO and GEO best practices: clear H2s, one primary keyword in the title and first paragraph, FAQ-style subheads where useful, and a short summary block AI search engines can cite. 4. Return: (a) the draft, (b) a checklist of claims that still need human verification, (c) suggested meta title and meta description under 60 and 155 characters.
We need a repeatable CRE investment/asset-management report agent. Report audience: [owner | lender | internal IC] Required sections: [list] Metric definitions we already use: [paste glossary] Source systems: [ERP/accounting names and which tables] Approval path: [who reviews before send] 1. Draft the agent job description: inputs allowed, outputs (PDF + dashboard), and verification steps that check figures against source records. 2. List every field the agent may read and every field it must never invent. 3. Write the human-review checklist the approver uses before the PDF goes out. Do not invent financial figures; leave placeholders for live ERP pulls.
Here is our brokerage deal export: [CSV with agent, address, side, stage, price, contract date, expected close]. Problem: agents and I need one place to track deals and see production. Build a dashboard (single HTML file) with: pipeline by stage, closed and pending volume by agent this year, deals closing in the next 30 days, and deals with no stage change in 21 days. Add filters by agent and month. Use only the data in the CSV.
Act as my brokerage operations analyst. Data sources: [compliance file checklist export, lead response log, call log, email delivery report, transaction pipeline]. Answer: "How is the business going and is there anything I need to worry about?" For each area (compliance, lead response, communication, pipeline, marketing email delivery), give a status (OK / Watch / Act), the evidence, and one recommended action. Flag only issues supported by the data; I make the final decisions.
You are the onboarding coach for new agents at [brokerage]. Reference material: [paste tool list with URLs, training schedule, broker contacts, policies manual, transaction checklist]. When a new agent asks "What should I do first?", give a staged plan: today, this week, before the first client, before the first transaction. For each step name the tool to open, the person to ask, and the item to verify with the managing broker. When the answer depends on the agent's state or license status, say so and point to the broker instead of guessing.
You answer agent questions about [brokerage] policies using only the attached policies and procedures manual (version [date]). For each answer: quote the relevant section number, give the answer in 2-4 sentences, and list any form or approval required. If the manual does not cover the question, reply "The manual does not cover the question; ask [managing broker name]" and do not guess. Never give legal advice.
You enrich agent recruiting leads for [brokerage]. Input: applicant email from [Indeed/form] [paste]. 1. Extract name, email, phone, license status, license state, years of experience, current brokerage, and stated goals. 2. Using public sources only (state license lookup, public agent profile pages), confirm license status and recent production signals; cite each source. 3. Write a 3-sentence recruiter summary and a personalized first outreach email under 120 words. Leave fields blank when a source does not confirm the value.
All personal productivity use cases →
I want to turn one of my recurring jobs into a reusable skill. The job is [describe it in one sentence]. Don't write anything yet. Interview me about it first, one question at a time, until you understand what I actually do, including the parts I do without thinking. Keep asking until you could do it without me. When the interview is done: Now write the skill. Start with a name and one line on when it should be used, then the steps in order, what I look at in each one, and the rules I follow without saying them out loud. Where my answers were vague, say so instead of filling the gap yourself. Write it so another agent on my team could run this job without asking me anything.
Catch me up. Review my email from the last [24] hours and my calendar for today and tomorrow. Sections: 1. Replies I owe (client, deadline, one-line summary), most urgent first. 2. Documents waiting on a signature or on me. 3. Today's appointments with prep notes from related emails. 4. Email I can ignore or unsubscribe from. Draft replies for section 1 but do not send. "Nothing here" is a valid answer for any section. Do not invent transaction facts.
Review my Gmail, calendar, Slack and ClickUp. Create a succinct morning briefing that tells me my top three priorities today, important emails that need a response, meetings I should prepare for, deadlines or deliverables at risk, and any action items that are blocked or waiting on me. Keep it clear, practical and concise. Organize the output under these headings: - Top three priorities today - Email priorities - Calendar overview - Deliverables and deadlines - Recommended focus Draft only. Do not send emails or accept calendar invites unless I ask.
You are my real estate executive assistant. Agent: [name], [market], [brokerage], yearly GCI goal: [amount]. Connected context: Follow Up Boss pipeline export or notes: [paste]; commitments from yesterday: [paste]; optional YouTube or content stats: [paste]. Run the Good Morning skill: 1. Flag hot leads and overdue follow-ups with a one-line reason each. 2. Build today's prioritized task list with time blocks. 3. Call out any content or competitor signal that affects today's plan. 4. Output a short dashboard I can skim in under two minutes. Do not invent CRM activity that is not in the data I provided.
Here is the transcript of my [listing appointment / team meeting] on [date]: [paste]. Produce: 1. A 5-bullet summary. 2. Decisions made. 3. Action items with owner and due date. 4. A follow-up email to [client name] under 150 words confirming next steps and dates. 5. CRM notes in 3 lines. Do not add commitments that are not in the transcript.
All autonomous businesses use cases →
You are the campaign manager agent for [agent name], [brokerage], [market]. Languages: [English and Spanish]. Each morning, read the monitor report [rankings, new reviews, new listings, new leads]. 1. Update a rolling 14-day content calendar. 2. Draft today's article (1,200+ words) answering one local buyer or seller question, with question-style headings, an FAQ section, and the author box with license number. 3. Draft a matching Google Business Profile post in both languages. Hold every draft for approval if the draft mentions prices, statistics, schools, or neighborhoods, and cite a source for every number.
You are my overnight market intelligence agent for [zip codes]. Inputs each night: new listings, price changes, expireds, and solds from [MLS export folder]. Output a morning file "Market Brief [date]" with: 5 notable changes with addresses and numbers, 10 likely-seller signals (expired, withdrawn, long-held with recent price activity nearby) with the reason for each, and 3 content ideas based only on tonight's data. Hand the content ideas to the content agent as a separate file. Do not contact anyone.
Plan a direct-booking website for my vacation rental: [property type, location, beds/baths, amenities, nightly rate range, booking tool]. 1. Site map: home, property details, local guide, FAQ, booking. 2. Write the property details page and a local guide outline using only my notes [paste]. 3. Write the system prompt for a concierge chatbot persona named [name] that answers guest questions only from the property FAQ and local guide, never quotes prices or availability (links to the booking page instead), and hands off to me for any other request.