One thing that has been touched upon is that you are setting the culture, if you say you’re hesitant to sacrifice velocity for tests already that will probably become the long term culture and one day there will be no tests and everyone will be afraid to make any changes. Tests do take up front investment but they run multiple times a day and thousands or even tens of thousands of times over the lifetime of a product and pay it back.
I think striking a balance where you’re not overly concerned about coverage but have key end-to-end flows covered can make things more efficient while saving your skin and help avoid unhappy situations. I say limited UI front-end tests (these are the greatest time sink) of key flows, but then as much coverage of apis as you can do.
Also I’m founder at Tesults (https://www.tesults.com). Check it out. There’s a free plan but I’ll go beyond that and give everyone who posted in this thread one full target free if you email me. So you can report your test results nicely too and keep track of failures without cost. Any direct feedback is appreciated.