Instagram began rolling out an AI assistant inside Edits on 30 September. Edits head Brett Westervelt told TechCrunch that it can use follows, views, video retention, likes and shares from a creator’s Instagram account together with comments, current trends and audience interests. The assistant looks for patterns across past posts and can suggest new ideas, hooks, captions, audio and scripts.
Meta had previewed the feature on 15 September as an assistant that could analyse Instagram insights and help brainstorm ideas. The initial rollout has usage limits, with Meta One subscriptions providing additional use. That makes the assistant a layer on top of data creators already receive from Instagram, rather than a separate measurement product.
YouTube is moving feedback into the draft stage
At Made on YouTube on 23 September, YouTube announced new Studio Insights and Research areas that use past channel performance and outlier-video research to help choose what to make next. It also introduced draft feedback on pacing, structure and storytelling, channel-matched thumbnail generation, dynamic thumbnails and background reviews of older videos that may benefit from new packaging. YouTube says testing up to three video cuts for different hooks is coming soon; its existing title-and-thumbnail A/B system has already been used in more than 40 million experiments since 2024.
First-party data becomes part of the creative loop
The shared shift is that platform analytics are becoming prescriptive before publication. Instagram can combine a creator’s own performance history with comments and platform trends; YouTube can compare drafts and packaging against channel history and research. Third-party tools may receive authorised analytics, but these assistants are built directly into the platforms’ creation and publishing workflows.
Neither company says these assistants expose the ranking algorithm or prove why one post succeeded. Their recommendations remain interpretations of observed performance and platform trends, and some YouTube features are still rolling out or forthcoming. The practical test is whether the advice improves future work across repeated posts, and whether lessons learned on one platform transfer to another.