A real estate agent in Riyadh today receives most of their inbound through three channels: WhatsApp, Instagram DMs, and the occasional phone call routed from a property listing on Aqar or Bayut. Email is for contracts, not inquiries.
That's a healthy signal — the customer is reaching out on the channel where they're already comfortable. The problem is what happens next.
The 11pm property inquiry
A buyer scrolls a listing at 11pm, taps "WhatsApp the agent", and types: "Is this villa in Al Yasmin still available? My budget is 2.4M, can do cash."
In the agencies we've talked to, three things happen to that message between 11pm and 9am the next morning:
- The agent on duty sees it on their phone, intends to reply, and forgets.
- The agent replies, but the next morning the buyer has already messaged two other agencies who replied faster.
- The agent's WhatsApp Web tab is open on a desktop that auto-locks; the message arrives, the agent doesn't, and by morning the inquiry is buried under 40 new ones.
None of these are bad agents. The pipeline is just leaking — and at a 2.4 million SAR average ticket size, a single missed inquiry per week is a serious annual number.
What "automation for property inquiries" actually means
The phrase "automation for real estate" gets used loosely. We mean something specific: a layer that sits between the agency's WhatsApp / Instagram / web channels and the agency's existing CRM (Salesforce, Odoo, Bitrix, or just a spreadsheet), and does three jobs.
1. Acknowledge instantly, in the right language. The buyer who sent that 11pm message gets a reply within seconds — confirming the listing is still available (or not), asking the qualifying questions a junior agent would ask, and promising a human follow-up at a specific time. Not "an agent will get back to you" — "Mohammed, your dedicated agent for Al Yasmin, will message you at 9am with photos and the floor plan."
The reply is in the buyer's language. If they wrote in Arabic, نظام الذكاء الاصطناعي replies in Arabic. If they wrote in English, English. We don't switch languages on the buyer mid-thread; that's a translation experience, not a customer experience.
2. Extract structured fields the moment the data arrives. The agent doesn't open a CRM at 11pm to log "buyer for Al Yasmin, budget 2.4M, cash". The automation does — name, budget, neighborhood, financing posture, urgency level, lead score 0-100. By 9am the next morning the senior agent's pipeline already shows the lead, sorted by hotness, ready to be worked.
The model also reads between the lines. "Cash" + "is it still available" + budget at exactly the listing price is a 90+ score. "What's the price?" on a property where the price is in the listing description is a 30 — they didn't read it, the conversion math is different.
3. Deduplicate across channels. The same buyer messages on WhatsApp at 11pm, Instagram at 9am the next day, and the agency's website widget at noon. Without dedup that's three new contacts in the CRM and three different agents working the same lead. With dedup, it's one contact, one conversation thread, one agent.
The dedup key is the phone number (normalized — +966555123456 matches 0555123456 matches 00966555123456). When the same number arrives on a different channel, نظام الذكاء الاصطناعي merges the contact and reassigns the conversations. The senior agent sees one row in their pipeline, with a small chip noting "3 channels".
What we're seeing in 2026
The agencies that have wired this up over the past 18 months report a few consistent patterns:
- First-response time drops from hours to seconds. The instant acknowledgment closes the most common loss path (buyer goes elsewhere because nobody replied).
- Pipeline conversion improves at the bottom of the funnel, not the top. Total leads stay roughly the same; what changes is the percentage of leads that the agent actually contacts within 24 hours. That's the real lever.
- Junior-agent role shifts. Less "follow up on overnight inquiries", more "qualify the warm leads the lead scored 70+". The senior agents work the 90+ scores. The cold leads get a polite "thank you, here are some other listings" auto-response and don't burn agent time.
- Compliance gets easier, not harder. Saudi PDPL requires a clear paper trail of who accessed what customer data. An AI layer that logs every read/write on every contact actually makes audits simpler than the WhatsApp-Web-on-shared-laptop status quo.
What you don't get
The automation doesn't replace agents. The agents we work with are emphatic about this — buyers want to talk to humans for the actual transaction. What نظام الذكاء الاصطناعي replaces is the pipeline-leak portion of the workflow: the part where messages get lost, leads go cold, and follow-up depends on someone remembering at the right hour.
Real estate is a relationship business. The automation's job is to make sure the relationship can start at 11pm if that's when the buyer feels like reaching out, so the agent can actually deepen it at 9am the next day.
What to look for if you're evaluating tools
A few questions worth asking any vendor in this space:
- Does it speak Arabic natively or via translation? If the latter, expect awkward replies. Native MSA is the bar.
- Does it deduplicate across channels? Cross-channel dedup is the unsexy feature that quietly delivers most of the operational gains. If a vendor doesn't have it, your CRM will fill with duplicates.
- Where does the data live? Under PDPL, customer data should be retained only as long as you have a live purpose for it. Your CRM is the canonical store; نظام الذكاء الاصطناعي layer should be a pipe that delivers structured data and lets its copy expire.
- Can you see the lead score's reasoning? A black-box "this lead is 87" is harder to trust than "this lead is 87 because they said 'cash', referenced the listing price exactly, and asked about the floor plan." Show your work.
The agencies winning Riyadh's mid-market in 2026 aren't necessarily the biggest. They're the ones whose pipeline doesn't fall asleep.