AI Unlocks Billions of Dollars for this One Group in Construction

AI's potential to reduce construction costs hinges on targeting those that hold the purse strings, not service providers.

One of the major themes we think about in the built world is housing affordability.  And much of the crisis comes down to construction — homes are unaffordable because building them is extremely expensive. 

There are hard costs and soft costs associated with construction, and ~20% of those costs fall in the “soft” bucket — think service work like design, architecture, structural engineering, legal, and financing. Design work, in particular, makes up a considerable chunk of soft costs. Solving for hard costs is very challenging (though we do believe next gen procurement companies can add value), but driving down soft costs with AI will unlock the next decade of real estate development and acquisitions for those who take the plunge  

When we first set out to explore this opportunity nearly a year ago, we thought this article would be about the possibility of AI for architects and engineers. We evaluated multiple AI co-pilots for service providers in AEC, including tools for architects, structural engineers, and even permit expeditors. 

However, after spending over 100 hours with practitioners, we now see the opportunity is in serving capital owners and developers who are truly incentivized to cut costs.

Example P&L of a Large Commercial Project

We believe the unlock to eliminating soft costs comes from leveraging Generative AI and computer vision to structure, understand, and improve design data — and one day even generating designs. This will significantly reduce soft costs, because, as it stands, design errors have major downstream effects on construction costs. 

After interviewing nearly 100 developers and unpacking the challenges in their projects here’s what we found:

• Over 60% of change orders are because of design errors and omissions.
• Change orders can cause 10-15% cost overruns in construction.
• The goal should be to use automation to get soft costs as close to zero as possible.

Most companies are approaching the problem wrong. 

They’re building AI co-pilots for each “spoke on the wheel” — co-pilots for architects, structural engineers, mechanical, electrical, and plumbing folks. In evaluating these companies, we discovered the following: 

• An incentive problem: The appetite for these AI co-pilots is low. The reason: a major misaligned incentive. Many of these service providers make money based on billable hours and have limited interest in taking on more work at lower fees. 
• A lack of data ownership: None of these service providers actually own the designs. In hiring architects, engineers, and consultants, developers are also buying and owning the data. If the goal is to acquire data to train a built world model, going after service providers is going to be an uphill battle.
• Challenging sales cycles: Sales cycles are too long, and there’s a general low willingness to pay for these AI co-pilots, resulting in challenging unit economics and limited TAM.

Our thesis  

Focus AI solutions on people who actually have incentives to lower construction costs. That means capital owners and developers. 

• With the industry going towards design-build — and designers moving in-house — there's a compelling opportunity to sell design software to developers. Historically, even products like Autodesk were pushed down by the Developers into the service providers. 
• We’re especially interested in tools that collect design data, which unlocks value for automating and improving designs over time. The largest collection of construction drawings in the world is up for grabs and incredibly valuable.
• Generating complete designs from day one is too hard right now and is likely not the fastest way to go-to-market
• The right wedge can unlock value engineering, collaboration software, subcontractor marketplaces, material purchasing and so much more
• We like companies who are leveraging AI to perform tasks 10X faster and >50% cheaper than the outsourced providers developers are using today. Especially if they have some layer of collaboration or workflow management bolted onto them with internal and external stakeholders!

Our POV: Builders in this space should analyze different design and service providers, assess how much value their design drives or service, and determine how easily it can be automated.

If you're building in this space and believe you have an opportunity to eliminate soft costs for developers, we’d love to hear from you. No matter what your approach is!

By focusing on these smaller, more manageable issues, we can make progress towards the larger goal of increasing housing affordability!

-Gabi and Jason