AI data ingestion creates opportunity for better Real Estate tools
AI has unlocked the ability for computers to understand meaning in a way similar to how humans understand. This removes the significant burden of data entry and maintenance faced by real estate developers.
.png)
Ask anyone in the real estate industry, and they are quick to tell you that the industry has not adopted technology. At Deco Base, we believe that artificial intelligence is going to be the killer feature that finally allows the real estate industry to effectively adopt software.
AI is going to be big for real estate. We interviewed over one hundred developers while researching this topic and this essay describes some of the key learnings that lead us to this conviction.
Data capture and entry is a huge pain point for real estate professionals.
A surprisingly common theme across all the conversations with real estate developers was that data entry was a huge barrier to using software. By definition, real estate exists in the real world. The necessary first step towards using software is a data entry step, in which the information that exists in the real world needs to be digitized. Unfortunately since each real estate asset is so unique, it is very difficult to build standardized data entry processes. There is always additional nuance that needs to be captured - everything from legal structures, to material choices, geographic position, market comps, architect, etc.
Computers have historically not been very good at capturing nuanced data in it’s full context. This means that if a developer does want to use better software, they first must first complete the data entry step manually. Even after entering data originally, if they wish their software to remain up to date with the latest information they must constantly update and refresh manual data inputs. The complexity of real estate assets translates to a huge number of variables that all need to be manually updated - eventually this simply becomes too much work for the average real estate developer. Ironically, the software that is supposed to make their job easier, actually makes their jobs more difficult.. An example of this is making updates to a proforma or a buyout log as new contract or invoice data is received.
Existing software for real estate developers is limited by bad data input.
Because it is so hard to wrangle all the context and data of a real estate development into a digital form, existing software is not able to deliver good user experiences. Existing software is extremely rigid, and even the most technologically savvy real estate developers become frustrated with maintaining high quality data. For example, a developer may use a service like bill.com to pay their invoices and have that feed into quickbooks which they then use to inform their pro-forma. The bill.com process works great, but the data pipeline breaks when the developer cuts a paper check to pay a pressing invoice on site. Bill.com has clean data, but it has limited context and can’t be used as the source of truth.
Another example is investment documents and return waterfalls. Real estate developers employ a variety of investment vehicles including equity, preferred equity, mezzanine debt, and construction loans. These combine into profit sharing and promote structures that can be somewhat complex. Existing cap table software is not equipped to model the detailed cash flows of a development project, meaning real estate developers often need to run custom scenarios when responding to diligence requests or preparing investor updates.
In general, the tech solutions that are available to real estate developers are very siloed or niche. Because real estate development projects are so nuanced, automatic data transfers are difficult for computers to execute. When developers do use that software, they still must must manually transfer that context or data to other tools. For example, using Docusign to collect signatures on investment agreements still requires that the terms and amounts be manually entered into a proforma or cap table.
Real Estate has scar tissue from tech that has over promise and under delivered.
There is a long history of proptech companies that have promised good solutions that have been unable to deliver. The fundamental reason that most companies have failed at this, is the data entry required to get their software to work is simply too burdensome. New products are created, but the fundamental workflows are rarely changed. The user interface might look more modern on new products, but it’s the same table structures and variables used by older products, which are the same tables and variables used in excel models.
When these tools try to talk to each other, they fail because the context and nuance is stored and communicated in different ways. Point solutions are shiny, but because of their challenges in interfacing with the rest of the real estate development process they are limited in usefulness.
Newer technology products have not been able to solve the issue of bad data inputs resulting in bad data outputs.
Artificial Intelligence will introduce new working processes to real estate.
One of the key abilities that language models have unlocked is the ability to structure unstructured data. In practice, what this means is that computers are now able to understand the world in a similar way to how humans understand the world. Computers are historically good at math, and bad at meaning. Language models give computers the ability to mathematically represent semantic meaning - which is a huge unlock.
For real estate, this means that the burden of data entry, transformation, and maintenance will no longer need to be done by a human. Instead of needing a human to read through an investment document and enter the key terms in a spreadsheet, AI based tools will be able to accurately do that on the human’s behalf. Computers have historically not been able to walk a jobsite with the developer and the GC. With computer vision models, they now can.
That’s not to say that AI will replace the need of Developers and GCs to be present on the jobsite, but it will replace the need of the developer to enter the info that they gathered from the job walk into a computer. Computers will now be able to operate with the full context that is available to humans.
Automatic data ingestion is going to create new opportunities for better real estate technology.
The increased ease of getting real estate context and data will allow for dramatically better software that will meaningfully improve real estate projects. The first and most meaningful impact that artificial intelligence will have on real estate is its ability to translate the unstructured real world into the structured, digital world of computers. This will drive a rapid increase in the quality of real estate software as the industry gains parity with other industries that have already digitized.
At Deco Base, we are working on the AI powered tools for real estate developers that will drive this change. Please reach out if you are interested in learning more. We’d love to hear your thoughts and ideas on AI in real estate.
