UpCodes Launches Copilot, an AI-Based Research Assistant for Building Codes

UpCodes He has been making the complicated world of building codes easier for building professionals and homeowners to understand. Its platform includes a searchable database of regulations for all states and functions as well as a "spell checker" that flags errors in the code. The startup announced a new AI-powered tool that will make navigating the world of building codes even more streamlined. Called Copilot and based on ChatGPT-4, Copilot acts as a research assistant, answering tricky code questions and annotating the answers with links to relevant sections of the code.

UpCodes has also announced that it has closed a Serie A of 3,5 million dollars, earmarked for recruitment as it continues to develop Copilot and add more AI-powered features to its platform. Coupled with previous UpCodes funding, including a pre-Series A announced in March 2021, this brings their total raised to $7,6 million.

The last round was led by Building Ventures, a venture capital firm focused on construction and real estate technology. Other participants include the co-founders of PlanGrid, CapitalX and Bragiel Bros.

UpCodes already has more than 650.000 monthly active users and more than 100 million page views. Scott Reynolds, its co-founder and CEO, said the startup's team has doubled, its revenue has quadrupled and its product offerings have expanded to include more user segments. The building code it covers has also grown from less than two million to more than five million housed sections, and now offers coverage for every state and major US cities.

Before launching Copilot, UpCodes focused on building its database of codes, often digitizing regulations that were only available in physical reference books, and making them easier to refer to. In addition to more than five million code sections, it is also home to 160.000 local amendments. The codes are constantly changing, so UpCodes updates about 7.000 per month on average.

Its database is searchable and has other tools designed to make code compliance easier, such as its code checking feature, but even using them is time consuming due to how complex the regulations are. Copilot aims to drastically simplify the code search process.

Reynolds gave some examples of the questions Copilot can answer:

• Calculate the travel distance for a specific occupancy in a building, or the maximum distance someone can travel in an emergency, to an exit (an example of Copilot's response to this type of query is included in the graphic above).
• Explain the context of a section of code to help understand its meaning.
• Look up related or more stringent code sections for other codes such as building, fire, and mechanical.
• Generate a residential roofing regulation checklist, with links to the relevant sections of the code.
• Copilot answers these questions and helps users by citing the sections of code from which it extracts information, so they can review the actual code for themselves.

“We've always been about education and helping users understand the underlying context,” says Reynolds. UpCodes plans to add more explanatory content to Copilot that helps users better understand the code, in addition to publicly available content.

Notably, laws hosted by UpCodes have contributed 0,01% of AI training data like ChatGPT and Google Bard. According to Reynolds, construction professionals often turn to these tools to answer their questions. They trained on the Common Crawl C4 dataset, pulled directly from the UpCodes website (basic access to their code database is free).

“Increasing attention is being paid to data quality in training LLMs,” he explained. UpCodes has an extensive library of high-quality construction laws, which is ideal for LLM training as we are the only online source for many of these laws.” It may not have been a conscious decision to include building laws, but rather the result of their algorithm identifying relevant quality data on a wide range of topics, including building laws."

But Reynolds added that the trace only includes snapshots in time, so his models are likely to be running from outdated code, as regulations are constantly changing.

This gives Copilot an advantage as it is based on the constantly updated database of UpCodes. In addition, the codes vary greatly from one jurisdiction to another. Reynolds says UpCodes has spent years building the infrastructure needed to keep its code database up to date.

One of the biggest challenges facing any AI project is limiting hallucinations.. Reynolds says that one of the biggest steps UpCodes has taken is to "fence" Copilot to the codes applicable to a project. This means he analyzes more than five million code sections, including 160.000 amendments, based on the user's location and year of permit. He also collects additional data, such as building type, to ensure that Copilot only uses the relevant codes.

Copilot has an in-house system that refines specific construction law responses and puts each query through multiple levels of analysis to better understand and analyze the questions. It focuses on jurisdiction-specific codes, to account for variances and because construction professionals often need to be familiar with multiple jurisdictions at once. Once a question has been asked, Copilot provides answers with context and direct references so that users can see how an answer was arrived at.

Users subscribed to an UpCodes payment plan can ask Copilot three questions. If they want unlimited access, they can upgrade to UpCodes Professional or add it to their Enterprise plan.

UpCodes Series A will be used to recruit your engineering department as well as all departments that contribute to Copilot. It plans to expand its code library and resources so that Copilot can produce more sophisticated responses and add new features, such as project management, according to Reynolds.

The startup team has known Building Ventures since its founding. “We think they would be the perfect partner for our Series A,” he added, explaining that the firm is made up of former operators in the construction sector, including former founders. Your portfolio and LP base serve as a valuable guidance resource when UpCodes need to validate ideas.

“The construction industry can be opaque, a world unto itself, so having investors and partners who have been in it for decades is helpful,” says Reynolds.

In a statement about Building Ventures' investment in UpCodes, partner Allen Preger said, “By unifying and maintaining all building codes on one AI-powered platform, UpCodes is transforming code compliance for the built environment. We are delighted to lead your Series A investment."