Buildings 2.0 Library
Conversations on the future of the built environment
81 episodes of the Buildings 2.0 podcast, each bound as a short book: the ideas, the numbers and the lines worth keeping.
Season 2, Episode 8 /Jul 3, 2025
Why engineers spend 50% of their time not engineering, and how AI fixes that
Daniel Calabro, Ferris
Jul 3, 2025 /AI & Automation
Why engineers spend 50% of their time not engineering, and how AI fixes that
Daniel Calabro is the founder and CEO of Ferris, an AI company for civil and structural engineers, and a structural engineer himself. He spent years watching engineers spend half their time on work that is not engineering. His interest in language models is practical: what they can safely do in a field where a wrong beam is a life-safety failure.
Daniel Calabro, Founder, CEO, Ferris
In this conversation
Daniel's explanation of what a large language model actually is, for engineers who use one every day without knowing.
The story of a new hire designing pre-tensioned concrete beams with a chatbot, and why the principal was right to be alarmed.
'Nobody verified layer one': his account of background drawings from major firms that were never checked, and why verification matters as much as prompting.
The UpCodes lawsuit and the question behind it: if code is law, should engineers have to pay to read it?
The judgment and the understanding of what's actually happening at a first principles level is always going to be on the human and on the engineer.
Daniel Calabro
Chapters
From engineer to founder
How a structural engineer decided to build software instead.
What a language model actually is
A plain-language explanation of the technology behind the chatbots.
The new hire designing beams with GPT
What happens when junior engineers start designing with AI unsupervised.
By the numbers
Construction workers who are 55 or older
Daniel says the most important data in a firm is the engineer retiring next month. The retirement wave is not hypothetical.
Source: Associated Builders and Contractors, via ENR (2025) ↗Nobody verified layer one
Why so much of the industry's base data was never verified.
Memory as a product
Ferris's thesis: project memory as the product.
The most important data within your firm is the engineer that's going to retire next month.
Daniel Calabro
Takeaways
AI should scale an engineer's mathematics the way analysis software did, while first-principles judgment stays human, the same way CAD replaced drafting boards without replacing design.
The most important data in a firm is the engineer retiring next month. Record their lessons and mistakes now, before there is a model to feed them into.
Building codes are treated as private documents even though the law holds engineers to them. Public, machine-readable codes would unleash tools the profession cannot yet imagine.
There's so many intelligent people working feverishly in our industry that if you just gave them more data, I can only imagine the tools they would build.
Daniel Calabro