AI's Impact on Dialogue Cleaning: RX vs. Newer Tools for Continuity?

The threshold for switching to ADR is directly impacted by the efficacy of AI-powered dialogue cleaning and matching tools, pushing that line further with each new advancement, making post-production sound more robust than ever. On my last feature, a documentary shot primarily on an AMIRA and using MKH 416s, we encountered a few challenging interviews where external noise bleed was significant due to less-than-ideal locations. Previously, a few of those takes would have been marked for ADR immediately.

Now, I find myself leaning heavily into Izotope RX 10, especially its Dialogue Isolate and De-reverb modules, for initial passes. However, I've also integrated newer AI tools like DaVinci Resolve's Voice Isolation and the Fairlight tools mentioned in the 'Dialogue Editing Workflow: Cleaning, Matching, and Continuity' guide (https://blockreeldao.com/blog/dialogue-editing-workflow-cleaning-matching-and-continuity) during the DIT and media management phase. I can often run preliminary clean-ups to mark questionable takes as 'yellow' instead of 'red' because these tools can salvage dialogue that would have been unusable just a few years ago. This shifts my continuity strategy because I'm spending less time worrying about perfect room tone matches, as these tools often create a more consistent sonic bed between takes than traditional methods. My goal is always to deliver the cleanest possible audio to the sound editor, minimizing their heavy lifting.

How are others adapting their continuity strategy, especially when it comes to preserving original performances versus calling for ADR, using these new AI-driven options?