Syed Imran Haider 
UpCodes is known for making complicated code building easier to understand for individuals, including homeowners and industry professionals. The platform comprises a database with all states’ regulations and features, including spell check to flag code errors.
The company has launched a new artificial intelligence-powered tool called Copilot that aims to streamline the navigation process of building codes. Built on OpenAI’s ChatGPT-4, the new feature will work as a research assistant, answer complicated code-related questions, and annotate responses using links to relevant code sections.
With over 650,000 monthly users, UpCodes has facilitated more than 100m page views. Besides the new tool, UpCodes announced the shutdown of a $3.5 million’s Series A, designed for recruitment, as it keeps building Copilot and introducing more AI-powered features for its platform. The company’s previous funding makes a total of $7.6 million.
Complicated Code BuildingÂ
A VC firm called Building Ventures led the recent round by focusing on construction and real estate technology. CapitalX, Bragiel Bros, and PlanGrid’s co-founders also participated in the venture. According to the CEO and co-founder of UpCodes, the startup has doubled the original team members now, as its returns have quadrupled and its products have included more user sections.
The company has expanded the building code from less than 2 million to more than 5 million hosted segments. UpCodes now provides coverage for all major cities and states in the USA. Before the launch of Copilot, the company built its code’s database and often digitized regulations accessible in reference books.
Apart from hosting more than 5 million code segments, UpCodes also accommodates 160,000 local amendments. It is worth mentioning that computer coding constantly changes, which makes the company update almost 7,000 on average every month. UpCodes has built a searchable database with some tools intended to ease code compliance similar to its code-checking-related feature, though the complexity of regulations makes it take lots of time. Copilot will vividly make the code research process simpler.
Examples of Answerable Questions
Mr. Reynolds provided the digital media with examples of queries Copilot could respond to. These include how to find out the distance one can travel in an emergency case or to exist, elaborating the idea behind a code segment to comprehend the meaning, or making a checklist for residential deck regulations using linked code sections. The new AI-powered tool can answer these questions and cite the code segments it extracts data from to help users, letting them evaluate the actual code.
According to Reynolds, the company has always leaned profoundly into education and helped users comprehend the context. UpCodes will add more descriptive content to Copilot to help users further understand code and publicly accessible content. UpCodes-hosted laws contributed to training data for ChatGPT and Google Bard-like AIs.
Reynolds says building professionals consult with those AI-powered tools to get their queries answered. They got training on the C4 dataset through Common Crawl pulled from UpCodes’s site. The extensive library of construction law of UpCodes is ideal for an LLM to train on as the only online source for lots of laws and regulations.
