YouTube Unveils Ask Studio, Dynamic Thumbnails and Video A/B Tools at Made on YouTube

YouTube used its annual Made on YouTube creator event on Wednesday to roll out a deeper suite of AI-assisted Studio tools aimed at helping channels break through, including Ask Studio feedback, dynamic thumbnails, multi-cut video testing, and live features due early next year, TechCrunch and The Verge reported.

Ask Studio lets creators share draft videos with YouTubes AI for personalized notes based on prior uploads and audience comments. The feature can propose alternative titles and thumbnails, explain performance using channel analytics, and suggest ways to improve metrics, TechCrunch reported. Ask Studio is also expanding to YouTubes mobile apps, widening access beyond desktop Studio workflows that already offered earlier chatbot-style analytics.

Dynamic Thumbnails allow creators to upload multiple images for a single video so YouTube can show different versions to different audience segments in pursuit of higher watch time. The Verge reported creators can upload up to three images per video, with the system assigning the strongest option by segment. YouTube already supported A/B tests for titles and thumbnails; the new layer leans harder on automated assignment rather than manual swap routines.

From 2027, creators will also be able to upload and test three different cuts of a Short or longer video—for example, alternate intros or structuresfed to small audience segments before a winner becomes permanent, TechCrunch and The Verge reported. The Verge said YouTube can auto-promote the winning cut after seven days if a creator does not choose manually, and that product leaders describe guardrails so the three versions are not dramatically different. Creators can also apply tools to older catalog videos, including autonomous thumbnail monitoring when permission is granted.

Amjad Hanif, YouTube’s vice president of creator products, told The Verge the background agent can watch for older videos that regain relevance to news or trends and suggest title or thumbnail tweaks, and can assemble demographic-backed brand pitches from channel data. Hanif drew a line between efficiency tools and AI that writes scripts or produces the content itself, arguing viewer trust frays when work no longer feels creator-authored.

On monetization and live, TechCrunch reported YouTube will use AI to auto-tag products for affiliate programs and is expanding its Amazon affiliate program beyond the United States. Early next year the company plans to pilot real-time auto-dubbing on live streams and launch live showdown matchups plus live creator matchmaking so co-hosts can compete for chats, gifts, and Super Chats, according to product manager Barbara MacDonalds blog notes cited by TechCrunch.

Platform data shared at the event underscored living-room viewing and short-form drama growth: more than half of the top 100 YouTube creators have living room as their most watched interface, TechCrunch reported, and microdramas surpassed 6.5 billion views in the first half of 2026 with watch time up 50% year over year and TV views of that format up 90%.

Skepticism remains part of the creator conversation. TechCrunch noted some creators worry a single AI mistake could damage credibility, while The Verge pointed to broader industry tension around AI research and production after high-profile creator controversies earlier this year. YouTube declined to share performance lift data when asked whether A/B tools and chatbots measurably raise views, Hanif telling The Verge the clearest near-term benefit is time returned to creative work.

For creators following platform product news on Influencers, the Made on YouTube package is less a single Icon Series” style drop and more a Studio workflow shift: testing, catalog refresh, live collaboration, and affiliate tagging now sit closer to the upload path. Whether Ask Studio and dynamic thumbnails become default habits or optional experiments will play out as the 2027 Shorts multi-cut tests and live dubbing pilots arriveand as audiences decide how much invisible optimization they still recognize as human-made.