The case
A bus that shows up is a public good.
If you finish this page and conclude you should buy something else, build it yourself, or keep the spreadsheet — good.
Start with who is waiting
This starts with a person on a bench at a country depot, twenty minutes early, because being late is not recoverable out here.
In a city, a missed run is an inconvenience. On a rural line there is no second option: the 6:15 is how you get to dialysis, the job, the pharmacy. When it does not come, nothing absorbs it.
The people most dependent on these services have the least capacity to route around their failure. Any account of what transit software is for that does not start there is measuring the wrong thing.
Why the market did not serve them
Not through malice, and not through incompetence. Through arithmetic.
Capable transit software is expensive to build. Large agencies can absorb five-figure implementation fees, so the sector's products were built for them — annual minimums tied to fleet size, a full-time professional scheduler assumed.
Each decision was rational. Together they left an operator running four vehicles and sixty route miles with nothing to buy, and so with a spreadsheet.
Roughly 2,200 transit operators exist in the United States and most are small. Whether they can run properly is not a niche question.
Four things we take as given
These are why particular things in the product are the way they are, and why we turn some business down.
01
Decisions belong close to the ground.
The person who knows why the Tuesday run leaves four minutes early is the person driving it. A template designed in another state just relocates that judgement away from the only person with the facts.
So the odd rule your operation has, for a reason, survives contact with our software. And where a statewide instance would serve you better than buying from us, our guides say so.
02
The work has dignity, and software either respects it or wastes it.
A mechanic who logs the same defect in three places has been told their time is the cheapest input in the system.
So nothing is entered twice, warnings arrive at sixty days rather than the morning of, and the driver's view is two taps because somebody is standing at a stop in the rain.
03
Nobody should be left carrying the whole thing alone.
In almost every small operation, one person holds the schedule, the history and the exceptions in their head. That is called a risk. It is also a burden nobody offered to share.
So notes attach to the line, the run and the vehicle rather than to a memory, and a second person can run a service change. Being indispensable should not mean being unable to take a week off.
04
You are stewarding something you were handed.
A depot built in 1911 stands because generations maintained it without expecting credit. The job is to hand it on in better condition.
So your data is exportable in full, at any moment, on every plan. A vendor that makes leaving difficult is a risk to the thing you are preserving.
What follows from that, in writing
Convictions that cost nothing are decoration. These cost us something.
| We will not | Because |
|---|---|
| Charge per trip | It taxes exactly the ridership growth agencies are being asked to produce, and makes grant budgeting guesswork. We charge per active vehicle instead, capped at the next plan's price. |
| Charge per seat | Per-person pricing is how drivers and volunteers lose access, and how the data goes back onto paper. Everyone at your agency gets a login. |
| Hide the price | A quote-gate exists to find out what you can pay. Our numbers are on the pricing page, including the per-vehicle charge and the onboarding fee. |
| Make leaving hard | Complete export, open formats, self-serve, every plan. Retention through friction is a confession. |
| Publish or send anything on its own | No automatic notices, no automatic messages to riders, no automation that contacts a human being without a human deciding. |
| Sell or train on your data | Your operational data is yours. It is not a second revenue stream and we will not make it one. |
| Take money we can't earn | If you run over fifty vehicles or need certified safety-critical integration, we will tell you to buy an enterprise suite. Our own comparison pages say where competitors win. |
On the question everyone is being asked right now
This industry has been sold a great deal. Transit OS is not an AI product and we will not describe it as one.
A language model does three small jobs here: condensing a headline, turning a feed-validation code into plain English, drafting a rider notice from an incident note. That is the list.
None of it acts on its own: a draft is published by a person or not at all. If the model is unavailable, the product loses a convenience and nothing else.
We think that is the right posture for software near a safety-critical operation, and we would rather be boring about it than fashionable.
Every figure we cite, and where it came from
If a number on this site is wrong, we would like to know. Tell us and we will correct it and say that we did.
- Rail transit leaders planning around budget constraints ahead of the September 2026 authorisation lapse — Progressive Railroading, Outlook 2026
- Scheduling and dispatching practice at small and rural agencies, including reliance on spreadsheets and the occasional-use training mismatch — N-CATT, Scheduling & Dispatching guidebook
- Counts of demand-response and vanpool systems — APTA Public Transportation Fact Book
- National operator counts and service summaries — FTA, National Transit Summaries and Trends
- 216 tourist railroads and museums operating roughly 2,734 route miles in North America — HeritageRail Alliance
- Third-party and supply-chain breach exposure against long-lived transit operational technology — Modeshift, trends shaping public transportation
- Fragmented, vendor-locked systems and tool sprawl reported by agencies of every size — Route Fifty; schedules.info
- Published enterprise demand-response pricing used in our comparisons — DemandTrans provider profile, Clean Mobility Options
If we have got something wrong, tell us.
We would rather hear that this misreads your operation than sell you something that does.