“
I solved NTD because I hated all the data entry.”
Chuck McCarthy, Transit Manager, Town of Winter Park, CO
Agency snapshot
Town of Winter Park, Colorado.
12 bus routes county-wide, 1-person back office.
Transit manager: Chuck McCarthy, CCTM.
The problem
Every transit agency in America owes the federal government an annual National Transit Database report. Chuck McCarthy became transit manager for the Town of Winter Park in 2023, having never filed one. His experience may sound familiar:
“It was basically like reading Klingon.”
“It’s kind of like doing your taxes for transit. It’s pretty ambiguous, just a bunch of forms and there’s no real step-by-step guide on what you do first.”
“It’s the least fun part of my job.”
At a KTS AI Launchpad workshop, Chuck stopped trying to do NTD the old way.
He rebuilt the process instead.
What he did · a method any agency can replicate
1
He gave Claude his real data
The master spreadsheet went in whole: financials, cost analysis, every trip, revenue and non-revenue vehicles, contractor costs.
2
He taught it the federal format
The layout of the forms, field by field.
3
He taught it how he had filed before
His past NTD reports — the original, revised, and final version of each year — so the AI learned not just the format but what the government had made him change, and why.
4
He had it build the workbook
Tabs mirror the federal forms closely enough that reporting comes down to copying values across. Yellow cells mark what needs entry.
5
He got ahead of the auditors
NTD flags data swings against validation rules it never publishes — so he reverse-engineered them from his own filing history into a built-in check:
pass
fail
warning
6
He built it for the people above him too
A final checklist tab, and what he calls the “CEO of everything” tab — the metrics a town manager asks for, populated automatically.
7
Then he built it again, for everyone else
A guided version for first-time reporters. A power-user version with the full formula chains. A help book and a quick-start guide. Creating the final took him four days.
“Before you submit to NTD, you can go through the workbook and know exactly what’s going to pass, what’s going to fail, and what’s going to be a warning. So you can address it before ever submitting to the portal.”
before AI adoption:
4 weeks
After AI Adoption:
with the workbook he built himself
What this means for you
“You don’t need to know code. You don’t even need to know how to get from A to Z. You just need to know where Z is.”
— Chuck McCarthy (not a programmer, and he will tell you so himself)
Take the task you dread most, the one you are sure is too tangled to fix. Describe where you want to land, hand AI your real work, and see how close it gets on the first try. Chuck’s worst job of the year now takes 25 minutes.
What’s your NTD?
The only thing between you and finding out is the question he asked: what would this look like if it just worked?
More stories
Transit
Resource Scheduling: Paddles
Turning thirty years of handwritten scheduling into a permanent system.
Case study
Read the story
⟶
Government
Regional Strategy: Streamlining CEDS
A five-county strategy that takes months, drafted live in one afternoon.
Case study
Read the story
⟶
Transit
Service Reliability: Predicting No-Shows
Taught a spreadsheet to see no-shows coming — six calls a night, seats kept.
Case study
Read the story
⟶