Artificial intelligence has been around for decades, proptech firms have been utilising it for at least five years, and real estate remains stubbornly resistant to adopting new technology.

So why should the sector get excited by the hype around generative AI?

It's 10 times better than anything that came before, Pi Labs Managing Partner and founder Faisal Butt said.

Butt dismissed fears there is an AI bubble and said a new generation of AI-based proptech firms is solving real problems for real estate firms and generating significant revenue doing so.

That wasn't the case a few years ago.

“A lot of the earlier AI startups had great technology, but you didn't really see the hockey stick curve in revenue growth in those startups,” he said.

“Now a lot of the startups that are being presented to us went from zero to a million revenue in the span of 12 months, and you’re like, wow. That’s the type of growth we want to see, but you're only seeing that because the technology is just completely superior.”

Specialist proptech venture capital investor Pi Labs has combed through thousands of startups, picked five it thinks will make it big and invited them to an annual accelerator programme it runs.

All of the companies use large language models to create their technology. Butt said the proliferation of AI real estate startups using the infrastructure created by LLMs is akin to the boom in app development following the invention of the iPhone. The trick is working out which have a proper use case and business model.

The companies in which Pi Labs has invested include San Francisco-based Fluix, which has developed an AI platform that acts like a facilities manager for data centres, helping optimise energy use and lower costs. Reducing energy usage by data centres is one of the key problems for the real estate sector today, Butt said.

Innex.ai, based in Cambridge, UK, has created an AI co-pilot that simplifies navigating governance documents, including legislation, regulations and standards for healthcare properties, focusing on the British National Health Service.

While that might seem very niche, the NHS is one of the world’s largest building owners, with a huge facilities management function, and healthcare more broadly is a massive global real estate asset class. So although it has a very specific function, the potential customer base is large.

Butt said investors are looking for companies like this that have a “defensible” position — a large potential customer base, but niche enough that the tech giants won’t try to muscle in.

“We're seeing some very, very narrow use cases that in and of themselves could be valuable businesses worth a couple of billion, but a couple of billion is not interesting enough for a GPT or Microsoft or Google to go after,” he said.

Another startup Pi Labs has backed is London-based Trubrics, a product analytics platform for LLM applications, such as AI assistants, AI co-pilots and voice commands. It allows businesses to better understand and increase the value of the AI products they build and launch by gathering insights on user sentiment and feedback.

Butt said the next stage of the evolution of AI that will impact real estate is the emergence of agentic platforms, which means AI systems that have agency and can make decisions themselves.

Pi Labs is evaluating an agentic AI startup for the property management sector, Butt said. The AI is not a software platform that you would subscribe to or a dashboard you would log in to — it's a virtual employee in your company that has an email address and a phone number and deals with customer inquiries.

“A year ago, I don't think GPT had inference,” Butt said. “Now it's got inference — every version of it is going to get smarter. So the smarter it gets, the more you move toward agentic AI and away from the AI co-pilot that just offers suggestions.”