Or have minibuses that are smaller and come more frequently. Find a few drivers willing to work after hours as a second job perhaps and let them drive smaller shuttle buses or larger passenger vans.

When big buses stop their routes the smaller ones would start.

The key is frequency and dependability.

Labor is a huge cost and these transit agency budgets are thin

this is a problem that autonomous driving could fix more easily than the general driving problem due to the fixed route, but good luck having the bus driver union not throw rocks at your AVs

I like the sibling comment idea of using AI eventually.

But in general for public transportation I think running at a loss is not unreasonable. It's nice if it's profitable, but it's ok if it isn't. Hopefully it's reducing emissions, giving people who can't or don't want to drive reliable access to their employment, etc. is still worth it.

the problem, generally speaking, is even if it reduces costs per ride, if it becomes popular it quickly becomes too expensive.

New York City tried replacing its dial-a-ride accessibility public transit service that required 24h advance appointments with an e-hailing subsidy and blew out its budget. https://www.thecity.nyc/2023/06/22/mta-paratransit-ehailing-...

Some places are experimenting with dial a ride shuttles