December 2023

Generative AI can change

real estate, but the
industry must change to
reap the benefits
Real estate can use generative AI to turn the industry’s
data into treasure in seven steps.

by Matt Fitzpatrick, Vaibhav Gujral, Ankit Kapoor, and Alex Wolkomir

Generative AI (gen AI) is maturing at an auspicious moment for the real estate industry.
Investors have mountains of both proprietary and third-party data about properties,
communities, tenants, and the market itself. This information can be used to customize
existing gen AI tools so that they can perform real estate–specific tasks, such as
identifying opportunities for investors at lightning speed, revolutionizing building and
interior design, creating marketing materials, and facilitating customer journeys while
opening up new revenue streams.

Although gen AI has only recently captured the public’s imagination, AI has been
fundamentally changing the way the world does business for decades. This more familiar
version of AI—also known as analytical AI—is goal oriented and focused on activities such
as predicting values for a future forecast or assigning categories to segment customers.
It is already embedded in parts of the business world: AI-assisted forecasts, for example,
have altered how investment professionals think about the future, and dynamic pricing
models have changed how several industries charge for goods and services. One
industry in which AI’s transformative power has been missing, however, is real estate, a
historically slow adopter of new technologies.1

Gen AI represents a fresh chance for the real estate industry to learn from its past and
transform itself into an industry at technology’s cutting edge. Gen AI has not replaced

1
“Construction in the cloud: An interview with Thomas Wolf, CEO of RIB Software,” McKinsey, April 30, 2018.
 analytical AI; instead, its open-ended and creative nature introduces a new frontier of
use cases that analytical AI does not address. Based on work by the McKinsey Global
Institute (MGI), we believe that gen AI could generate $110 billion to $180 billion or more in
value for the real estate industry.2

For all the hype that gen AI has received to date, many real estate organizations are
finding it difficult to implement and scale use cases, and thus have not yet seen the
promised value creation. This is not surprising: deriving competitive advantage from
gen AI is not as simple as just deploying one of the major foundational models, and many
things have to go right in an organization to make the most of the opportunity. To help
companies get started, this article will explore the technology’s potential and provide
examples of use cases, each of which real estate actors we work with are in some phase
of exploring or implementing. The article will then describe seven pivotal, interconnected
actions that real estate companies can take to realize the full value of this gen AI–fueled
vision and create strategic distance from their peers.

Real estate can benefit from gen AI in a multitude

of use cases
Gen AI’s strengths generally fall within one of four categories, which we dub “the four Cs.”
The first is customer engagement, which can be supported by tools such as conversational
chatbots that answer questions and remove doubt from customer decisions. The second
is creation, in the form of tools that generate new creative content, including text and
images. The third is concision: gen AI excels at synthesizing insights from unstructured data,
interpreting conversations, and querying large data sources. The fourth is coding solutions,
of which gen AI offers many, including interpreting, translating, and generating code.

In our own work with AI, we have seen real estate companies gain over 10 percent or more
in net operating income through more efficient operating models, stronger customer
experience, tenant retention, new revenue streams, and smarter asset selection. Here are
five examples of how businesses can apply gen AI’s four Cs to specific real estate issues.

Sifting through mountains of leasing documentation (concision)

Gen AI can be applied to a repository of lease documents, which can be dense and filled
with bespoke terminology, making it difficult for owners of many properties to sift through
and find information at scale. A gen AI–powered tool can summarize key themes across the
leases, such as how much rent is expected monthly or what market forces (such as local
environmental, social, and governance compliance laws) could affect leases. Additionally,
the tool can scan across leases for a particular parameter (for example, all leases with a rent
price per square foot below a certain level) and generate tables of information. At that point,
professionals can examine the information the AI tool has compiled.

Copiloting real estate interactions (concision and customer engagement)

Gen AI can be used to create a powerful copilot (a gen AI–powered bot) for a variety
of real estate interactions, including managing tenant requests and lease negotiation.

2
The economic potential of generative AI: The next productivity frontier, McKinsey, June 14, 2023.

2
 Simple requests from tenants, such as for routine maintenance, can prompt the copilot to
directly contact a building’s maintenance staff. The copilot can identify a more complex
question and flag it for a specialist at a property management company. As the specialist
interacts with tenants, gen AI can observe conversations and written responses
and suggest ways to improve communication. For high-stakes moments—such as a
commercial lease negotiation with an office, warehouse, or retail tenant—a gen AI tool
can take in all the information about a tenant, the property, and the market and craft a
negotiation transcript. If communications and calls are recorded or turned to text, the
copilot can monitor these interactions at scale, providing coaching while reminding
specialists to refrain from using certain terms that could incite moments of risk.3

Enabling visualization and creating new revenue streams (creation and

customer engagement)
Today, when a prospective office tenant looks at raw space on a tour or a potential
resident views pictures of an apartment on a listing site, they see an empty unit or photos
filled with someone else’s finishes and furniture. Virtual reality tours have helped, but
these static, noncustomizable simulations usually only go part of the way toward showing
the end user what the result could be.

Gen AI tools can help a potential tenant visualize exactly what an apartment would look
like in, say, their preferred midcentury modern style or in cherrywood versus walnut
finishes. This data can then be fed back into a model to predict which types of furnishings
and finishes work best for different customer segments, improving prospect-to-lease
conversion and shaping future capital expenditure decisions.

There can also be e-commerce tie-ins: as a prospective tenant tours a unit, an app can
virtually impose a variety of couches, window trims, or kitchen appliances that match a
desired design style. If the prospective resident decides to buy or lease, these choices
can be ordered and set up to coincide with the move-in. The resident benefits by moving
into a home that already expresses their signature style, and the brokerage or apartment
company benefits by reaping revenue from cross-selling.

One large furniture retailer has launched a gen AI–powered product visualization tool
that enables users to upload a photo of a room and populate it with furniture from its
catalog. A variety of businesses throughout the value chain can use this capability to
create new revenue streams while deepening customer loyalty.

Making faster, more precise investment decisions (concision)

Today, investment decisions are often informed through individual analysis of bespoke
data pulls across sources. An investor interested in warehouses, for example, typically
starts by performing a macroanalysis of markets that have attractive factors such as
ports, airport locations, and high e-commerce volume. Then, they perform more granular
analysis to locate areas of interest, pulling building information from local brokers or
digital tools. As part of the decision-making process, the investor conducts discrete
analyses to figure out how their investment hypotheses have panned out in the past.

3
Nico Grant and Karen Weise, “Microsoft and Google unveil AI tools for businesses,” New York Times, March 16, 2023.

3
 With a gen AI tool that’s fine-tuned using internal and third-party data, an investor can
simply ask, “What are the top 25 warehouse properties up for sale that I should invest
in?” or, “Which malls are most likely to thrive in the future?” The tool can sort through
the unstructured data—both internal (such as the performance of a company’s existing
properties and the lease terms related to this performance) and third party (such as the
US Census and publicly recorded, comparable sales). This multifaceted analysis can be
overlaid on a list of properties for sale to identify and prioritize specific assets that are
worth manual investigation.

Drawing architectural plans known to create desired outcomes (creation)

In website design, there are specific patterns and design choices known to generate
e-commerce sales or higher click-through. Similarly, there are underlying design
principles in the physical world that gen AI can unlock and use to draw architectural plans.

A gen AI–assisted process can introduce Internet of Things sensors and computer vision4
algorithms that collect data points on space use, such as how customers move through
a store before purchase or when conference rooms are used in an office. This insight—
along with outcome data about sales, customer loyalty, productivity, employee retention,
or other areas—can then be fed to a gen AI tool. This information can be overlaid with
spatial data about square footage, location, walls, furniture, and other architectural
elements. The gen AI tool can then develop architectural plans that are optimized to
create desired outcomes in a space. Human architects and designers can work from
these plans to ensure art and emotion in the design, but with less guesswork over
whether a space is purpose driven (illustration).

Seven pivotal actions real estate players can take to

realize the full value of gen AI
Gen AI holds the promise of transformation, but real estate companies will have to do more
than just learn how to use off-the-shelf products. Although foundational models are essential,
they are just a small component of a real estate firm’s ability to realize value from gen AI.

To seize the opportunity, businesses in the real estate value chain can strive to
outcompete by rewiring the way they work in the following ways.

Align the C-suite around a business-led road map tied to a specific part of the real
estate value chain
CEOs who want to lead in gen AI can prioritize technology, onboard new internal
capabilities, and organize for agile delivery just as top start-ups and tech-native
companies do. New ways of delivering technology are essential not just to gen AI delivery
but also to ensuring modernity and staying ahead of the strategic curve. Winners are
willing to experiment, iterate, and self-disrupt.

That starts with having capabilities that go beyond the traditional real estate IT
organization. This does not mean leaders have to welcome scores of new tech hires

4
 subfield of machine learning that uses visual data, such as images, videos, and 3-D signals, extracting complex information and
A
gaining rich interpretations.

4
 Daylight

Foot traffic

Noise

Illustration of gen AI–assisted architectural

designs that show where daylight would fall on
this design, how the plan would accommodate
typical foot traffic patterns, and whether noise
levels would distract people in the space.

Source: McKinsey research and experience with

sensor and computer vision data
5
 into their companies. Rather, it requires investing in a nimble squad of engineers and
designers who are familiar with gen AI and can be directed to focus exclusively on value-
adding use cases.

C-suites can start by assessing which part of the real estate value chain they occupy—
such as development, operations, or investment—and considering how the journeys of
tenants, employees, and other stakeholders can be reinvented. Then, they can begin
redesigning roles and structures to make the alignment happen. Getting value from gen
AI requires that executives be willing to question the industry’s traditional hierarchies
and operating models and, most important, to accept a new technology layer throughout
the organization. Gen AI requires executive-led adoption of new ways of working that will
elevate the power of professionals across functions and levels.

Adopt a laser focus on data—especially new proprietary data

In the gen AI future, those with access to and control over unique, informative data will
be able to generate insights that others cannot. Companies can start by thinking about
what data they need—as well as what proprietary data about tenants and properties is
available but not currently being collected.

It is essential not only to have the best data set but also to have it engineered the right
way with the right data governance. A conversational AI tool that has been trained on a
building’s past maintenance requests can efficiently respond to resident complaints. A
tool trained on a real estate portfolio’s net-operating-income data can provide answers
about performance that could be useful for investment decisions and for reporting to
investors and internal company divisions. Internet of Things sensors and computer vision
applications in office buildings, for example, can provide anonymized insights into how
tenants use spaces, creating nuanced views of the built environment.5 Tenant apps and
dashboards are not merely interaction channels; they themselves can become data
sources. What kind of amenity space a residential tenant books, what stores a shopper
in a mall browses, or what services an office tenant needs to produce an event are all
valuable pieces of data that can be harnessed and structured.

It is important that this harnessing and structuring occurs in a data lakehouse controlled
by the real estate firm (as opposed to by a third-party system). A variety of vendors
should be able to interact with this single source of truth. This will allow real estate firms
to cut data by building, tenant, or type of unit or space for their own internal uses and
to be flexible enough to switch systems and applications easily. Companies should be
conscious of data ownership and should make the ability to easily access and work with
data a key part of vendor decision making as they design their future tech stacks.

Engineer a prompt library that gets results from foundational models in a real
estate context
Foundational gen AI models are only as good as the questions (known as “prompts”)
asked of them. As models are fine-tuned with data specific to real estate, it’s important to
engineer a prompt library.

5
Craig Smith, “Can sensor technology help keep office workers healthy,” New York Times, November 23, 2022.

6
 A prompt could say, for example, “Use the following resident history and property data
to craft an initial outreach email to a resident looking to renew their lease,” followed by,
“Generate a follow-up email based on their response and consider offering one of the
following more-personalized concessions based on what you know about the resident:
waived pet fee for two months, complimentary deep carpet cleaning, or membership to
the building’s gym for six months.”

Slight edits in syntax, detail, or framing can yield meaningfully different outputs with an
impact that can only be discovered in action. There is no precedent for knowing what
works until it is tried. To create a playbook, a rigorous process of testing and refining to
ensure questions return expected answers is essential.

Create digital tools that promote action—not just insight

Newly popular large language models fascinate the public because of how easy it can be
to write prompts and receive a comprehensible response. But real estate firms should
not be lulled into thinking that all gen AI outputs are intuitive or that plug-and-play point
solutions deliver the full promise of gen AI. Instead, outputs often need another layer of
digital tooling to be useful to an organization.

A foundational model, for example, may generate marketing copy, but it may require
additional tooling to check it for grammar and brand compliance and ensure the language
abides by regulations. A gen AI model may come up with insights on how customer
service should be delivered by a real estate agent, but agents may need to receive
prompts in a specific cadence during client interactions or explanations that clarify why a
particular recommendation is important.

Design is another crucial component. Color, style, and physical-design patterns for how
a user clicks are paramount in traditional apps. With a gen AI interface, it could be more
important to fine-tune a conversational system’s writing style or to make sure the customer-
facing avatar and its spoken tone and pitch are agreeable to the audience and encourage
desired behaviors. In short, the definition of design will have to expand as a new playing
field of psychology tied to interactions with algorithms and machines emerges.

A self-service portal that makes it easy for employees to access company-approved tools
and learn how to use them can enable organization-wide innovation.

Invest in a modern technology stack to enable data use

A vital leg of the gen AI stool is the tech stack: the right infrastructure, feedback loops,
safeguards, and integration should be built in a secure, scalable, and user-friendly way.

Gen AI requires new capabilities relative to traditional AI and machine learning,

including toxicity checks (ensuring gen AI is not creating problematic content, such as
words or statements that would violate relevant fair-housing or antidiscrimination laws)
and guardrails against hallucination (preventing gen AI from providing false answers
without sharing that the tool is uncertain). Compared with traditional data science
and analytics, gen AI relies more on engineering unique tech stack elements to make
it effectively operational—capabilities that real estate businesses may lack in their
current IT organizations.

7
 Real estate companies across the value chain that embrace early proofs of concept and
start orienting their tech stacks in the right direction to enable use cases in the future will
be in the best position to benefit. Fruitful actions include taking the time to thoughtfully
link vendor systems and connect the dots on data across property management systems,
customer relation management, and maintenance portals.

Adopt a new operating model that can scale as a real estate portfolio grows
and diversifies
To enable a gen AI upgrade to processes for investing, leasing, and more, operating
models and jobs may need to be redrawn to match the new focus points of work.

New roles and capabilities, such as prompt and data engineers who can implement
foundational models, may be needed. People in existing roles, such as agents or on-site
staff, may be able to hand off time-consuming tasks to gen AI tools, allowing them to
focus on specialized tasks. In other cases, such as on marketing and investing teams, gen
AI may evolve the discipline and create demand for new roles and skill sets. Companies
need to be open to change, because the face of the IT or marketing organization will not
look the same with AI tooling, even if the objectives of the business unit remain the same.

Recognize and mitigate risks unique to real estate created by gen AI

Gen AI is new, maturing, and facing open questions. There may be biases in training
data that are unintentional but create outputs with real consequences. There may also
be questions about the intellectual properties feeding foundational models as the legal
precedent around the space evolves.6 Marketing content, for example, may emerge from
an algorithm trained on unlicensed images, catching the real estate business anawares.
Provider tweaks to the foundational model that underpins a real estate–specific tool
could create unforeseen consequences for output quality and content, making it
important to reengineer prompts.

Additionally, some use cases may not be right for gen AI in its current stage. These might
include emergency response, where the stakes are too high, or leasing decisions, where
it could be difficult to comply with regulations that require explanation.

Deploying gen AI in situations where investment professionals, operating teams, and

tenants meet introduces new risks to working, living, and shopping environments
that have to be weighed. Companies that identify risks early on and iterate to find
improvements will be positioned to react effectively.

We believe that the time is right for real estate to make the leap to gen AI for two
primary reasons: first, the technology has now caught up to the problems that real
estate companies face and, when properly designed and executed, can provide game-
changing solutions. And second, commercial real estate today faces headwinds, and
these new challenges demand innovative approaches. Gen AI will not replace analytical
AI—there are use cases (such as producing a rent forecast or a retention prediction) for

6
Christopher Mims, “AI tech enables industrial-scale intellectual-property theft, say critics,” Wall Street Journal, February 4, 2023.

8
 which more traditional machine learning excels. Rather, gen AI is opening up use cases
that were never before possible and are relevant to dimensions of the real estate value
chain that technology did not previously touch.

But can an industry that is a reputed tech laggard overcome structural challenges,
invest in underlying technology, restructure operating models, and emerge as a gen
AI leader? The fact that real estate has been a laggard may actually be an advantage,
because the industry did not overinvest in previously available but now-outdated
technology as some other industries did.7 The learning curve may be steeper for those
who have not yet invested in technology and data strategies, but we believe gen AI’s
insights, speed, and transformational power for the real estate industry are worth the
climb. Data—and use of that data to create strategic distance—are the new terrain on
which businesses will compete for investor dollars, tenants, buyers, and longevity.

The idea of getting started can be daunting, but we urge executives to start simply.
Our technology professionals advocate for a “2x2” approach: identify two use cases
that can launch a company into taking ownership of data, deliver measurable impact
quickly, and build excitement; and identify two use cases that are more aspirational,
will fundamentally change the business, and take more time to deliver. This approach
encourages companies to push the technology toward its full potential.

There is no question that the real estate industry is changing. The question is which
companies will answer that call of change, adapt, and embrace a new, digitally powered future.

7
“Should US banks be moving to next-generation core banking platforms?,” McKinsey, July 26, 2022.

Matt Fitzpatrick and Vaibhav Gujral are senior partners in McKinsey’s New York office, where
Ankit Kapoor and Alex Wolkomir are associate partners.

The authors wish to thank Anannya Das, Dan Eversen, Dan Guo, Kat Huang, Ivan Serov, Rahul Sharma,
Dmitry Shatalin, Jason Smith, Annabelle Tang, and Nicolette Tran for their contributions to this article.

Copyright © 2023 McKinsey & Company. All rights reserved.