How Real Estate Companies Use AI Today: Real Examples from JLL to Bayut

Nearly every large property company is now experimenting with AI, and almost none of them say it is fully working. JLL surveyed more than 1,500 senior commercial real estate decision makers and found 88% of investors, owners and landlords piloting AI, yet only 5% had achieved all of their program goals.

What the data shows

• 88% of investors, owners and landlords and 92% of corporate occupiers are running AI pilots, but only 5% report achieving all their goals (JLL, October 2025).
• More than three quarters of commercial property professionals now report using AI, with valuation, market analysis and portfolio decision support as the top uses (RICS, August 2026, 3,148 responses).
• 72% of facility managers say AI already affects their work every day (IFMA 2026 Global Salary and Compensation Report).
• 20% of US Realtors use AI tools daily, and ChatGPT is the tool used by 58% of those who use AI (National Association of Realtors, September 2025).
• Real estate firms plan to spend less than 1% of revenue on AI in 2026, against about 2% for tech companies and financial institutions (BCG AI Radar 2026).

So adoption is wide but shallow. The useful question for a developer, brokerage or property manager is not whether to use AI, but which jobs it is already doing well for others. Below are the real examples, grouped by the work they change.

Valuation and market analysis: AI estimates built on official data

Valuation ranks among the top AI uses in commercial property, according to RICS, and it is where the Gulf has some of the clearest examples.

In June 2024, Bayut launched TruEstimate, a property valuation tool built with data integration from Dubai Land Department. It estimates a property's current market value using recent sales of similar properties, market trends and property characteristics. Users can enter a Title Deed, Oqood contract, unit number or DEWA number to generate a valuation report.

At the large advisory firms, AI mostly works behind the scenes on research. JLL introduced JLL GPT in August 2023 as an internal assistant for its workforce, and by October 2024 more than 47,000 JLL professionals had used it. JLL then launched JLL Falcon, a platform that combines JLL's own property data with generative AI to produce insights for clients. In March 2026, JLL said about a quarter of its 110,000 employees use its internal AI platforms every day, mostly to research deals, build cash flow forecasts and develop applications.

The lesson from both examples is the same. The AI is only as good as the transaction data and research underneath it. Bayut's estimates lean on official DLD records; JLL's lean on its own research and property data.

Lead handling and property search have moved into chat

Buyers and tenants increasingly start their search by typing a question, not by setting filters.

On 6 October 2025, Zillow became the first real estate company with an app inside ChatGPT. People can ask questions like "What can I afford nearby?" or "Show me homes with a big backyard," see listings with photos, maps and prices, then move to Zillow to book a tour or contact an agent.

The UAE followed quickly. On 1 June 2026, Bayut became the first UAE based real estate platform to launch a property search app on ChatGPT, letting users describe location, budget, size, bedrooms and property type in plain language.

Property Finder's December 2025 white paper explains why agencies care. Between 2022 and 2025, the number of active agents grew 30% a year and listings rose 34%, yet engagement per listing fell 36%. Just 5% of agents now generate more than 40% of platform leads. Its AI supported Credit Optimizer tool reported 26% more leads on upgraded listings, 30% lower cost per lead, and up to four hours saved each week for agency admin teams.

Brokerages are also putting AI directly in the hands of agents. Compass, the largest US residential brokerage by sales volume, launched a voice activated assistant in June 2025 that drafts emails, creates follow ups and builds listing presentations. In July 2026 it added AI Assistant, which delivers daily briefings, flags "Likely to Sell" opportunities, updates client records and drafts follow up emails using context from each agent's contacts and deals.

Individual agents are doing the same on their own. The National Association of Realtors found 46% of members use AI generated content, such as listing descriptions.

Document review and lease work: from days to minutes

Real estate runs on long documents: leases, sale contracts, service charge statements and title records. This is where some of the biggest time savings are showing up.

Cushman & Wakefield worked with AI company Unframe to automate lease abstraction, the job of pulling key terms out of a lease. As reported in August 2026, processing time dropped from up to three days to minutes, and the work has grown to more than 15 active AI projects across Cushman & Wakefield teams.

Regulators are moving too. In August 2026, Dubai Land Department announced it would showcase Malik, its round the clock omnichannel customer service assistant, which DLD lists as powered by OpenAI, alongside the Dubai REST app as the single access point for real estate services.

Leasing, maintenance and property management

The most measurable results so far come from operations, where the same questions and repairs repeat thousands of times.

A survey of 280 executives at multifamily operators, published by leasing AI company EliseAI in October 2025, found 68% had integrated AI into existing business systems. Among operators using AI, 85% reported measurable improvements in lead to lease conversion, 76% reported faster maintenance resolution and 77% reported moderate to significant reductions in operating expenses. Notably, 78% said they had lost new business to competitors using AI. Because EliseAI sells these tools, treat the numbers as directional.

CBRE offers a longer track record. In August 2023, it reported its AI based Smart Facilities Management solutions were running at more than 20,000 client sites covering 1 billion square feet, reducing maintenance costs and energy use by as much as 20% and cutting technician dispatches by an average of 25%.

In the UAE, Aldar and Visa completed a real service charge payment in December 2025 in which an AI agent in the Live Aldar app confirmed the customer's details and paid on their behalf.

Why most real estate AI projects stall, and what the leaders do differently

If AI is this useful, why have only 5% of firms in JLL's survey hit all their goals? The research points to the same cause again and again: the information underneath.

JLL's January 2026 guide found 49% of real estate leaders name data workflows as their top AI use case, yet only 40% have the data foundation to support it. RICS found commercial property's leading barriers are data quality and availability and regulatory uncertainty. And BCG's spending numbers suggest the sector is underinvesting compared with others.

The companies with real results share three habits:

1. They start with one high volume, document heavy job. Lease abstraction at Cushman & Wakefield, maintenance dispatch at CBRE, listing performance at Property Finder. Each has a clear before and after.
2. They feed AI their own records. Bayut uses DLD transaction data. JLL uses its own research. Generic AI without company information produces generic answers.
3. They control who sees what. Property firms hold owner details, commission terms, tenant files and unreleased prices. An AI that every employee can query has to respect the same access rules as the files themselves, or it becomes the easiest way to leak them.

Questions people ask

Which real estate companies use AI the most?

Among named firms, JLL is one of the most advanced, with about a quarter of its 110,000 staff using internal AI daily. Zillow, Compass and Bayut have built AI into search and agent tools, and CBRE has run AI in facilities management across 1 billion square feet since 2023.

Can AI value a property accurately in Dubai?

AI tools like Bayut's TruEstimate produce estimates from recent comparable sales and DLD transaction data, which makes them useful for a quick view of value. They are estimates, not formal valuations. For lending, disputes or investment decisions, a qualified valuer is still required.

Where should a mid sized brokerage or developer start?

Start with the work that repeats most: answering buyer and tenant enquiries, drafting listing descriptions and follow ups, and pulling key terms from contracts. Train the team on those tasks first, then connect the AI to your own listings and documents so the answers reflect your business.

CroVIN sets up private company AI for property firms so every agent's AI knows your listings, contracts and processes, and only what each person is allowed to see.