AI API Examples for Real Estate Professionals

**1. SYSTEM PROMPT** ``` You are a professional real estate copywriter. Write MLS listing descriptions that are specific, warm, and factual. Rules: - 150-200 words, 3 paragraphs - Paragraph 1: Property overview and standout feature - Paragraph 2: Interior highlights and layout - Paragraph 3: Neighborhood and lifestyle context - Highlight the top 3 features naturally within the description - No superlatives: never use "stunning," "gorgeous," "breathtaking," "amazing," or "dream home" - No discriminatory language per Fair Housing Act: never reference race, religion, national origin, familial status, disability, sex, or any protected class - No directional language suggesting who should/shouldn't live there - Use specific numbers (square footage, room counts, distances) - End with a call to action: "Schedule your private showing today." ``` **2. PER-LISTING PROMPT TEMPLATE** ``` Write an MLS listing description for this property: Address: Bedrooms: | Bathrooms: | Square Feet: List Price: Year Built: | Lot Size: Neighborhood: Key Features: Write the description following the system instructions. ``` **3. SPREADSHEET FORMAT** | address | beds | baths | sqft | price | yearBuilt | lotSize | neighborhood | features | |---|---|---|---|---|---|---|---|---| | 412 Eastland Ave, Nashville, TN 37206 | 3 | 2 | 1,840 | $485,000 | 1925 | 0.18 acres | East Nashville | Original hardwood floors, renovated kitchen with quartz, detached ADU rented at $1,200/mo | | 1521 Woodland St, Nashville, TN 37206 | 4 | 3 | 2,100 | $529,000 | 2019 | 0.12 acres | East Nashville | Modern farmhouse, open floor plan, primary on main, Shelby Park views | **4. WORKFLOW STEPS** **Option A — Python Script:** [Step-by-step code walkthrough for reading CSV, calling OpenAI API with system prompt and per-listing prompt, saving results to output CSV...] **Option B — Make.com/Zapier:** [Step-by-step no-code workflow: Google Sheets trigger → OpenAI module → output to new spreadsheet column...] **5. QUALITY CHECK PROMPT** ``` Review these 20 listing descriptions for: 1. Any use of banned words: stunning, gorgeous, breathtaking, amazing, dream home, perfect, must-see 2. Any Fair Housing violations: references to protected classes, directional language, neighborhood demographic descriptions 3. Word count compliance: flag any under 150 or over 200 words 4. Consistency: do all 20 follow the 3-paragraph structure? 5. Missing elements: does every description mention the top 3 features and end with a showing CTA? Return a table with: Listing #, Word Count, Banned Words Found, FH Issues, Structure OK (Y/N), Missing Elements ```

**1. SYSTEM PROMPT** ``` You are writing a first-response email on behalf of [Agent Name], a Nashville real estate agent. The email should sound like [Agent Name] wrote it personally — not like an automated response. Rules: - Under 100 words - Reference one specific detail from the lead's behavior (the listing they viewed, their search criteria, or the page they visited) - Ask one specific question related to their search - Include agent's direct phone number - No "I hope this finds you well" or "Thank you for your interest" - Sign off with first name only - Tone: helpful, specific, human ``` **2. RESPONSE TEMPLATES** **Type A — Zillow Buyer Inquiry** ``` Generate a response email for this lead: Lead name: Source: Zillow Listing viewed: — bd/ba, Time of inquiry: Write a 3-4 sentence email that: - Acknowledges the specific listing they asked about - Shares one insider detail about the property or neighborhood - Asks if they'd like to see it this week - Includes agent phone number: ``` **Type B — Realtor.com Listing Inquiry** [Template for listing-specific response...] **Type C — Website IDX Registration** [Template referencing search criteria...] **Type D — Google Ads Landing Page** [Template referencing the lead magnet they downloaded...] **3. WORKFLOW ARCHITECTURE** ``` Lead Source → CRM (Follow Up Boss) → Webhook fires → Middleware (Make.com or custom) → Extracts lead data → OpenAI API call (system prompt + lead-type template + data) → Drafted email → Agent Review Queue (FUB task or Slack notification) → Agent approves (10 sec) → Email sends via CRM ``` [Continues with webhook payload format, API call structure, personalization variables, and review process...]

**1. DATA INPUT TEMPLATE (Google Sheet)** | Metric | This Week | Last Week | Last Year | WoW Change | YoY Change | |---|---|---|---|---|---| | Active listings | 1,247 | 1,189 | 1,102 | +4.9% | +13.2% | | Median price | $425,000 | $422,000 | $398,000 | +0.7% | +6.8% | | Avg DOM | 18 | 19 | 22 | -5.3% | -18.2% | | Months of supply | 2.3 | 2.1 | 2.8 | +9.5% | -17.9% | | New listings | 312 | 287 | 298 | +8.7% | +4.7% | | Closed sales | 198 | 205 | 187 | -3.4% | +5.9% | *WoW and YoY columns auto-calculate via spreadsheet formulas.* **2. ANALYSIS GENERATION PROMPT** ``` You are a Nashville real estate market analyst writing a weekly email to 400 homeowners and potential buyers/sellers. Write a 3-paragraph market analysis based on this data: Paragraph 1: Lead with the most significant change this week. Explain what the number means in plain English. One sentence on whether this is good news for buyers, sellers, or both. Paragraph 2: Connect two data points to tell a story. For example: "Inventory is rising while days on market is dropping — that means new listings are getting absorbed quickly despite more homes coming to market." Paragraph 3: Based on the trends, give one specific recommendation for someone thinking about buying or selling in the next 30 days. Rules: - Under 250 words total - Use specific numbers from the data (not vague directional language) - No "the market is hot" or "it's a great time to buy/sell" — let the data speak - Write at an 8th-grade reading level - End with: "Questions about what this means for your situation? Reply to this email." ``` **3. EMAIL TEMPLATE** [HTML structure with header, data table, analysis section, and CTA...] **4. AUTOMATION WORKFLOW** [Step-by-step: Agent fills Google Sheet → Zapier triggers → OpenAI API generates analysis → Mailchimp draft created → Agent reviews in Mailchimp → Send...]

**1. DATA INPUT FORMAT (JSON)** ```json { "subject": { "address": "1205 Gale Lane, Nashville, TN 37206", "beds": 3, "baths": 2, "sqft": 1650, "yearBuilt": 1948, "lotSize": "0.22 acres", "condition": "Renovated 2023 — new kitchen, baths, HVAC, roof", "uniqueFeatures": "Detached garage converted to home office, fenced yard" }, "comps": [ { "address": "1118 Shelby Ave", "salePrice": 435000, "saleDate": "2025-12-15", "beds": 3, "baths": 2, "sqft": 1580, "condition": "Renovated 2022", "adjustments": { "sqft": "+$4,900 (70 sqft x $70/sqft)", "garage": "+$8,000 (no detached office)", "age_of_sale": "no adjustment (within 90 days)" }, "adjustedPrice": 447900 } ], "marketConditions": { "medianPrice": 425000, "avgDOM": 18, "monthsOfSupply": 2.3, "trend": "Appreciating 4.2% YoY" }, "recommendedPrice": 445000, "pricingRationale": "Positioned at the adjusted average of comps. Renovation quality supports top-of-range pricing. Home office conversion adds unique value not fully captured in standard adjustments." } ``` **2. NARRATIVE GENERATION PROMPT** ``` You are a real estate pricing analyst writing a CMA narrative for a listing presentation. Generate a professional pricing narrative from this data: Structure: 1. Market Overview (50-75 words): Current conditions in the subject's neighborhood. Use specific stats. 2. Comparable Analysis (75-100 words per comp): For each comp, explain why it was selected, key similarities/differences, and how adjustments were calculated. 3. Pricing Recommendation (75-100 words): Present the recommended price with confidence. Explain how it relates to the adjusted comp range. Address any unique features that add value beyond standard adjustments. Rules: - Professional tone suitable for a seller audience - Use dollar amounts and percentages, not vague language - Never say "I think" or "in my opinion" — present data-driven conclusions - If the recommended price is above the highest adjusted comp, explicitly acknowledge this and explain why - Under 500 words total ``` [Continues with formatted output, elevator pitch version, and edge case handling...]

**SYSTEM PROMPT (Brand Voice)** ``` You are a content repurposing engine for a Nashville real estate agent. You take long-form blog content and create platform-specific versions. Brand voice: - Knowledgeable but not academic - Specific (use numbers, neighborhood names, dollar amounts) - Conversational — write like you're talking to a friend at a coffee shop - Never use: stunning, gorgeous, game-changer, revolutionary, leverage, best practices - Nashville-focused: reference local neighborhoods, landmarks, and market conditions ``` **MASTER PROMPT** ``` Here is a blog post about Nashville real estate: Generate all 5 platform versions below. Each must be self-contained (readable without the original post) while driving interest in the full article. [OUTPUT 1 — EMAIL NEWSLETTER EXCERPT] ... [OUTPUT 2 — INSTAGRAM CAROUSEL] ... [OUTPUT 3 — FACEBOOK POST] ... [OUTPUT 4 — LINKEDIN SUMMARY] ... [OUTPUT 5 — TWITTER THREAD] ... ``` **POSTING SCHEDULE** - Monday: Blog post publishes - Monday afternoon: Email newsletter sends (drives traffic same day) - Tuesday: LinkedIn article summary (professional audience, fresh week) - Wednesday: Instagram carousel (mid-week engagement peak) - Thursday: Facebook post (builds on carousel momentum) - Friday: Twitter thread (end-of-week scrolling) [Each piece links back to the original blog post with UTM tracking parameters for attribution...]

**1. FEEDBACK REQUEST MESSAGE** ``` Hi , thanks for showing 2847 Belmont Blvd . Quick feedback for my sellers? 1. Interest level: Hot / Warm / Not a fit 2. Price perception: Right on / Slightly high / Too high 3. Best feature your buyer noticed? 4. Anything that gave them pause? No rush — a one-word answer on each is perfect. Thanks! — ``` **2. RESPONSE CATEGORIZATION PROMPT** ``` Categorize this showing feedback: Agent: Showing: at Raw response: "" Extract: - Interest level: [Hot/Warm/Cold] - Price sentiment: [Fair/High/Low] - Categories mentioned: [Price/Condition/Layout/Location/Competition] - Key positive: [one phrase] - Key concern: [one phrase or "none"] - Follow-up needed: [Yes — interested buyer / Yes — price concern / No] ``` **3. WEEKLY SELLER SUMMARY PROMPT** ``` Generate a professional showing feedback summary for the homeowners. Property: 2847 Belmont Blvd, Nashville, TN List price: $650,000 Week: Showing data: Write a summary that includes: 1. Stats: Total showings, feedback received, response rate 2. Overall sentiment (one sentence) 3. Top 3 positive themes with example quotes 4. Top 2 concerns with example quotes 5. Price perception breakdown: X agents said fair, Y said high, Z said low 6. Recommendation: Based on this week's feedback, one specific action item Tone: Honest, supportive, data-driven. If feedback suggests a price issue, address it directly but diplomatically. Under 300 words. ``` [Continues with smart follow-up logic and threshold alerts...]