Constructed-world AI start-ups lead investments overtake FinTech and advertising and marketing

(© Andrey_Popov – Shutterstock)

Generative AI models can draw the most attention in areas like marketing automation and financial technology. But a recent analysis by A/O PropTech reports that investments in AI startups are growing faster in the built world, both in terms of volume and value of investments. These investments have experienced a significant upward trend in the course of preparing for new sustainability mandates.

As the name suggests, A/O PropTech is a venture capital investment firm specializing in the built environment. At the moment, AI is still in its infancy in the built world due to the prevalence of unstructured data that is more difficult to calibrate and reconcile than other domains. An additional caveat is that A/O PropTech has a broad view of the built world, including real estate management and insurance of real assets that others may put into financial services.

Climate technology is the fastest growing AI application including climate risk, ESG reporting and energy management. The largest deals are concentrated in the most mature segments, including real estate transactions, property management, and construction.

Getting beyond AI laundry

Catriona Hyland, research analyst at A/O PropTech, said they initiated the latest study to see how the recent hype around generative AI is impacting startups in the built world. So far, generative AI has not caught on in the industry due to the numerous challenges in making sense of unstructured data. In the short term, generative AI could play a much more important role in expanding datasets to provide a foundation for other tools.

She said they were trying to isolate some of the AI-laundering issues that had plagued previous research on AI startups. A 2019 study by MMC found that 40% of startups reportedly using AI weren’t. Further analysis revealed that this was mainly due to third parties touting the AI ​​capabilities, which the companies had never claimed or corrected. Her team developed a large language model to parse data from Pitchbook and Crunchbase for analysis. Hyland explained:

I think AI is often used as a kind of buzzword when you have a startup where AI is the core component of that company. But I think in the built world, and probably most industries, it’s used more as a tool than a centerpiece of sorts.

London conducts business

The study found that AI-enabled built-world startups in Europe and North America have raised $18.6 billion in venture capital over the past decade, almost half of it in the last two years ($8.6 billion Dollar). And in both 2020 and 2022, venture deals in AI-enabled Built World startups overtook FinTech AI funding, reaching over 600 deals worldwide in 2021 alone.

Also, London saw the most deals while the San Francisco Bay Area saw more capital. London also recorded more deals than Paris, Berlin, Dublin and Tel Aviv combined. When asked why Hyland explained:

When we looked at the breakdown of the types of transactions that took place in London, many of them were at a fairly early stage and much more focused on areas like real estate transactions and the financial aspects of the built world. This makes sense given London’s position within Europe as a sort of financial hub.

Structure the unstructured

In the short term, they expect a greater focus on using AI to understand structured data rather than generating new building designs. Innovations in computer vision are proving incredibly important for scanning job sites, tracking progress and automating insurance claims processing. But other types of generative AI are still in their infancy.

Jess Clemans, Investor at A/O PropTech stated:

The innovation around the big language models and image generation models has been pretty much hyped in the media, but we haven’t seen much of this filter in real estate technology. I think we’re close to using a lot of it, but it’s still early days. What we’ve seen is a lot of other innovative AI coming into built world technology, but we’re interested in seeing where this is going and trying to get a better picture of how we think these are doing new innovations in the built world will be used next-generation technology.

The challenge is that buildings fundamentally obey the laws of physics and have a lot of regulations and logic about how they are structured and how they need to be built. You can’t leave it entirely to a machine to determine the output. We’ve seen companies use this approach. And the result has often been very illogically attached bathrooms to kitchens, hallways that lead nowhere, or windows that don’t connect to bedrooms.

What we’ve seen as a slightly more successful approach in this area is mixing generative AI models with human-readable rule systems that the AI ​​must adhere to. So I think we’re going to see some sort of merging of algorithmic approaches with generative AI approaches in space.

A big challenge is that developers are still trying to figure out how to specify things like building processes and building codes in a way that the AI ​​can understand. Today, most construction data is stored in paper files or PDF documents. Significant efforts are required to catch up. In addition, building codes can vary between cities and countries. Technical and subjective aspects must be taken into account, which local governments approve.

Clemans explained:

There is no actual structured rule system. This is so in its infancy that each company that has done this has invented their own system for assigning rules to items. And it’s specific in general. So we’ve seen companies do this for electrical or plumbing system design. And they had to apply it a little differently than someone who does architectural floor plans or architectural detail designs.

But the last piece of the puzzle that’s really interesting and complicated is basic human logic. For example, if you decide to place an electric railing in the ceiling, you may not want to attach it directly to the edge of the wall because the mounting screws are difficult for a human hand to reach. But a machine would never understand that. This is just an example of some of the finer details that are not part of building codes, that are not part of typical building documents, but part of human logic that needs to be translated in that direction in the future.

My recording

The generative aspects of AI are getting all the press this year. But better data translation and matching can offer more value in the short term. Improving workflows and processes for building and operating physical infrastructure requires finding better ways to make good use of data collected for other reasons.

The recent crop of generative AI applications emerged from Google’s efforts to develop a better translator between French and English. These transformer models could also be important in translating documents, designs and 3D data collection into digital twins.

The UK has been one of the innovators in BIM (Building Information Modeling) technology for organizing data about the built environment. It will be interesting to see how this lead plays into the UK government’s 10-year plan to make the UK a global AI superpower.