If you read nothing else on this topic, read the DORA Report (pdf)
An idea I first came across in the Increment article "Testing the boundaries of collaboration" by Kent Beck. Kent is one of the founders of extreme programming and and original signer of the Agile Manifesto (which is often, er, poorly implemented these days).
At their core, feature flags are if statements. The core benefit, however, is that the extension of this pattern allows us to merge to our main branches more often and separate the deployment of code from the release of that code to users.
let FEATURE_A = true;if (FEATURE_A) {// do_something()} else {// do_something_slightly_different()}
Honeycomb, the current leader in o11y practices, has started talking about something called [[Observability Driven Development]]. The integrates well with the feature-flag driven deployment world.
Observability, IMO, is a continuation of the DevOps thread to this point: An engineer should own their changes from development through to production, and ensuring that their changes did what was intended.
This demo shown by Ellen Chisa shows off a way of working with software that isn't widely adopted yet.