TL;DR:
- Artificial intelligence (AI) is being increasingly utilised in the real estate industry to improve efficiency, enhance decision-making processes and offer personalised experiences to buyers and sellers.
- More and more startups and corporates are developing new value propositions along this value chain.
- We at PT1 believe that AI in real estate is revolutionary! PT1 is investing into more transformative real estate technologies, and we stand ready to support founders who are building in this space.
Is there value in incorporating AI in real estate?
This is probably a question every investor in the built environment has been asking themselves. We currently observe many companies developing or acquiring AI solutions that promise to provide faster and more intelligent insights to both their organisations and clients. Management companies are also embracing „as a service“ business models, including software, which has become possible due to the cloud and data transformation that optimises AI utilisation. A prominent example is JLL, which recently introduced its own GPT model specifically designed for commercial real estate. Instead of a simple chatbot, JLL GPT exemplifies the company’s data-centric approach to the industry’s digital transformation.
The rapid advancements in AI have resulted in the emergence of large language models (LLMs) and new forms of behavioural interactions on the Internet. Real estate professionals are leveraging language learning model technology to revolutionise the industry and improve user experience, which enables automated property descriptions, intelligent buyer-seller matchmaking, and more accurate market trend predictions. In this article, we will explore some of the ways AI is used throughout the lifecycle of a building. But, before delving into specific real estate and construction use cases, it’s important to understand the broader applications of LLM solutions that have already seen significant adoption.
Large language models and their applications in the current market
A large language model (LLM) is a type of artificial intelligence (AI) algorithm that uses deep learning techniques and massively large data sets to understand, summarise, generate and predict new content. The term generative AI also is closely connected with LLMs, which are, in fact, a type of generative AI that has been specifically architected to help generate text-based content.
A common solution offered by LLM is enhanced customer experience. This is mainly done through automated customer service chatbots, task automations, recommenders and improving information conciseness. Companies like Dialpad, Intercom and PolyAI are successfully operating in this space and using AI to transform their customer experience.
Another heavy use case of LLM is in generating creative content for marketing messages, press releases and image generation. In some cases, AI can even create songs and videos. For example, Jasper and TextCortex help companies realise their personal intelligence and help them create, communicate and collaborate in an effective way. However, there are some intellectual property concerns that limit companies from generating multimedia applications, though text-based applications are rapidly increasing.
LLMs are also used for assisting in coding through accelerating engineering productivity and developing innovative approaches to previously challenging tasks. For example, Github Copilot and Tabnine cater to coders and developers. Those who use these products code up to 55% faster.
AI in real estate: Is it too early?
AI in real estate is not too early – It is already being increasingly utilised in the real estate industry to improve efficiency, enhance decision-making processes and offer personalised experiences to buyers and sellers.
Property Search and Analysis: AI-powered platforms can gather and analyse massive amounts of data, including property features, historical sales data, market trends to provide accurate property valuations, pricing recommendations and investment analysis. For example, Plunk delivers a comprehensive AI-driven financial analysis of property, helping stakeholders monetise residential real estate investments. They also provide predictive analytics and market trends such as property price fluctuations and demand.
Virtual Assistants and Chatbots: AI-powered chatbots are being employed by real estate firms to provide instant responses to customer inquiries, assist in property searches and schedule property viewings or appointments. Additionally, virtual assistants can offer personalised recommendations based on customer preferences and frequently asked questions.
Image Recognition and Virtual Staging: AI can analyse property images to identify objects, rooms and attributes automatically. This technology enables virtual staging, where empty or unfurnished spaces are virtually furnished, allowing potential buyers to visualise a property’s potential. It can also highlight damages, structural issues or other concerns. Matterport offers virtual 3D virtual tours to properties saving time and money for all stakeholders involved.
Risk Assessment and Fraud Detection: AI algorithms can assess potential risks associated with property purchases or rentals, evaluating factors like crime rates, flood zones and nearby amenities. Additionally, AI can detect fraud and prevent fraudulent listings or financial transactions by analysing data patterns and user behaviour.
Smart Buildings and Energy Efficiency: AI can enhance the management of commercial buildings, optimising energy consumption, detecting maintenance issues and predicting equipment failure. This technology helps improve sustainability and reduce operational costs by shifting demand response and improving IAQ. Solutions from startups such as Simplifa and Hank improve predictive maintenance of buildings, optimise efficiency in processes and lower running costs.
Value of investing in PropTech-specific startups and how we vet startups in the current market
As prominent investors in the European built environment technology ecosystem, we evaluate whether it is wise to invest in a PropTech-specific LLM startup, considering the growing popularity of general LLM tools like ChatGPT and Google Bard. We are certainly extremely interested in meeting PropTech and ConstructionTech-specific AI solutions. However, specific industry LLM tool seeking funding from us will need to address the following questions:
- What unique proprietary data does your solution possess, which sets it apart from other tools, and what value does this data bring to its output?
- How is this data accessed, and how do you ensure its legality and compliance?
- What distinct value does this industry-specific solution offer that a general tool cannot?
Before the real estate industry fully embraces LLM, there is significant room for real estate companies to become more data-driven in their organisational processes and these companies will need to address the following:
- Strategically organise and manage the data they have access to, establishing clear data pipelines.
- Centralise data collection.
- Visualise the data effectively.
- Develop analytical routines that empower professionals to regularly analyse their data.
Several PropTech firms already excel in providing these services for asset management, property management and individual building management. LLM can enhance accessibility and submission of data in a streamlined manner, benefiting individuals across the organisation who may be unfamiliar with new technology or resistant to managing multiple logins and passwords. Ultimately, LLM has the potential to improve the frictionless experience of accessing and submitting data.
Output risks associated with LLM models
While LLM can generate quick reports with clarity, there are several risks associated with its output that should be considered:
Fairness and Bias: Due to imperfect training data or decisions made by engineers during model development, LLM may generate algorithmic bias.
Intellectual Property (IP): Both the training data and model outputs can pose risks to intellectual property, such as infringing on copyrighted, trademarked, patented or otherwise legally protected materials. Organisations must understand the data used in training and its implications for tool outputs, even when utilising a provider’s generative AI tool.
Privacy: Concerns regarding privacy may arise if user input ends up in model outputs in a way that allows individuals to be identified. Additionally, generative AI can be exploited to create and spread malicious content, including disinformation, prejudice and hate speech.
Security: Adversaries can utilise generative AI to enhance the sophistication and speed of cyberattacks. The technology can also be manipulated to produce malicious outputs, as demonstrated by the technique known as prompt injection where a third party provides new instructions to the model resulting in unintended outcomes for the producer and end user.
Traceability and Explainability: Generative AI operates through neural networks with billions of parameters, making it difficult to trace and understand the pathways through which outputs are produced.
Reliability: LLM models may provide different answers for the same prompts, making it challenging for users to assess the accuracy and reliability of the outputs.
Call to action: Revolutionising AI in real estate together
AI in real estate can be deemed as too early for some or just a trend every other industry is adapting. But we at PT1 believe that AI in real estate is revolutionary. PT1 is investing into more transformative real estate technologies, and we stand ready to support founders who are building in this space. We also encourage you to connect with us as we embark on this transformational journey together!
