Strategy for differentiating transient errors from bugs

We run a Django application which reports errors to Sentry. One issue we have is how to differentiate between transient errors (e.g. external service goes down) and errors caused by bugs.

Because of the nature of some of our deployments we get a lot of transient errors which makes it easy to miss errors that are caused by actual bugs.

Does anyone have a suggestion for how to deal with this?