01 · Industrial operations
Built for a client
The margin nobody trusted.
A pricing question should not trigger another round of spreadsheet archaeology.
Do not use AI to guess what the business can calculate.
Read the full storyStories from the work
How four familiar problems became useful AI systems.
See the original work, the system we built, and how a person checks the result.
These examples are based on real work. We remove names and identifying project details unless the client has given us permission to share them. Work still being delivered and examples using sample data are clearly labeled.
02 · Civil and environmental engineering
Being built now
The process estimate that took too long to explain.
The answer was not in one file. Neither was the reasoning behind it.
A trustworthy system makes disagreement visible.
Read the full story03 · Advisors and insurance
Built for a client
The carrier PDF that slowed down every client meeting.
The advisor knew how to guide the client. First, someone had to find and check the numbers.
For important work, every answer needs a way back to the source.
Read the full story04 · Innovation and patents
Built for a client
The patent search that was hard to check.
A confident answer was not enough. The company needed the evidence behind it.
Good AI research preserves both the source and the uncertainty.
Read the full storyYour first useful problem
What work inside your business deserves a closer look?
Bring the tasks that waste time, cause mistakes, or depend too heavily on one person. We will compare them and explain which opportunity is worth testing first.