In a two-part episode of The Future Built Smarter podcast, IMEG software development team lead Steve Germano discusses IMEG’s internal, AI-powered chatbot, Meg. After a year-and-a-half of development, Meg is now live and serves as a search engine for the firm’s engineering teams and other departments, drawing from the company’s vast amount of stored data.  

Built as a large language model, or LLM, questions can be asked of Meg by anyone in the company on a variety of engineering and non-engineering topics, from “How do I submit my expense report?” to “Where can I find guidance on sizing steam traps?” Meg will then point the user to the appropriate in-house tool or information from among the firm’s own curated, accurate, and verified databases. Meg fields upwards of 1,000 questions a day from the firm’s 80-plus engineering teams and other staff around the country, quickly bringing knowledge from across the firm to answer questions and provide technical information to provide the best solutions for local clients.  

Steve Germano

“We built a singular entry point to help users find information they’re looking for across the company in a single place,” says Germano, who also is a mechanical engineer. “Everyone can just type a sentence and ask what they want, just like you’d be asking a colleague sitting next to you.”  

Far from replacing engineers, Meg acts as an assistant to help them more quickly find the data and answers to their questions—an especially useful “co-pilot” and source of accelerated learning for less experienced engineers (who also continue to be mentored by the firm’s veterans). 

“It’s like having someone you can bug and ask 50 questions a day and know you’re not going to aggravate them and eat up their time,” says Germano, who expects to see similar AI-powered assistants to be developed across the AEC industry. “As the technology continues to develop, it’s just going to get better and better, and more and more knowledge will be available.” 

Germano offers a bit of advice for firms thinking about embarking on such a journey. “There are a lot of tools out there to start exploring with, but in parallel with that, you need to determine if your data is even ready to be consumed by AI. That’s a deep topic that needs to be explored as it can take a lot of time to curate and cleanse your data.” 

Listen to the podcasts: