Field notes
Short notes from the work: where capacity actually hides, how to size an opportunity, and why most AI initiatives in AEC stall before they reach a workflow.
There are hundreds of things AI could do inside a 60-person engineering firm. Maybe eight are worth doing this year. The difference between a firm that gets value and one that doesn't is rarely the tooling — it's whether someone sized each opportunity by hours consumed, frequency, and how hard it would be to change the workflow around it.
An RFP arrives, someone reads it, extracts requirements, hunts for a similar past project, pulls resumes, rewrites boilerplate, routes for review, and formats. Almost none of that is judgment work. The judgment is go/no-go and the win theme. Structuring approved content so a first draft can be assembled is usually the single highest-leverage change a firm can make.
A firm where twelve people have ChatGPT accounts has adopted a tool. A firm where the submittal log routes itself and the monthly report assembles itself has redesigned a workflow. Only the second one shows up in utilization.
If an engagement ends and nobody inside the firm can identify the next opportunity, the work didn't land. Teaching a PM or operations lead to recognize repetitive, rules-based work and scope an automation for it is worth more than any single automation we build.
Next step
An assessment maps your workflows, sizes the opportunity in hours, and prioritizes what to build first.
We intentionally limit the number of active engagements so we can stay deeply involved in the work.
