YouTube adds draft feedback and thumbnail tools to Studio
Creators are getting more help before and after publication, with mobile access to Ask Studio and further video-testing features planned.
The briefing
YouTube adds tools to review unpublished videos and test their presentation. Google outlines private memory for assistants across devices, while NVIDIA introduces power and cooling qualifications for AI data centres.
YouTube adds draft feedback and thumbnail tools to Studio
YouTube is expanding Studio with AI feedback on unpublished videos, generated titles and thumbnails, and mobile access to Ask Studio. Tests of alternative opening cuts and automatic thumbnail changes are planned.
YouTube’s new draft-feedback feature reviews unpublished videos and suggests changes to pacing, structure and storytelling. Ivan Mehta reports for TechCrunch that the company announced the additions at its Made On YouTube event.
Ask Studio, its AI question-and-answer tool, is expanding to iOS and Android. A new research feed will also show creators which kinds of content are working on the platform, giving them material to consider when planning videos.
Testing the packaging, then the opening
Studio will let creators generate titles and thumbnails from a video’s content, matched to their style. A dynamic-thumbnail feature produces three options and shows them to different audience segments.
YouTube says creators have already run more than 40 million title and thumbnail A/B tests. It plans to extend testing to as many as three video cuts, allowing creators to compare opening hooks.
Automatic monitoring and replacement of poorly performing thumbnails is planned for later in 2026. YouTube also says its updated analytics will explain performance and suggest improvements, rather than simply display numbers. Those explanations remain AI-generated advice, not proof that a suggested edit will win viewers.
Creators can bring draft review, presentation choices and audience testing into the same channel-management app. The planned opening-cut tests would let them compare actual edits, rather than judge a video solely by its title and thumbnail.
Market signal
Google outlines encrypted cloud memory for assistants across devices
Google has detailed a persistent-memory architecture for Private AI Compute. It aims to let assistants retain context across devices while protecting stored information through device-held keys, isolated processing and software verification.
Private AI Compute has so far discarded context when a task ends. Google now plans to add encrypted, per-user storage so an assistant can carry information from one request or device to another.
In its technical update, the Google Private AI Compute Team describes a system in which personal devices hold the keys to that storage. Google claims the design would keep the information inaccessible to anyone else, including Google itself.
Memory inside an isolated environment
When an assistant needs stored context, an authenticated, end-to-end encrypted connection would link the device to a secure cloud enclave. The enclave would temporarily decrypt the information in isolated memory, process the request, then encrypt any updated context for storage.
Google gives examples such as retrieving assembly instructions previously viewed through smart glasses or continuing a conversation between mobile and web. These are intended uses of the architecture, not an announcement that those experiences are available.
Alongside an updated technical paper, Google is publishing what it calls a tamper-proof public record of server software. Devices would check that software before sending personal data. The company also says it is sharing independent audit results, opening its privacy claims to outside scrutiny.
An assistant that remembers across devices needs somewhere to keep personal context. This design would move that storage into the cloud while leaving access dependent on users’ devices, rather than relying only on the provider’s account controls.
What to watch
NVIDIA launches power and cooling qualification for AI data centres
NVIDIA’s DSX Ready programme qualifies selected power and cooling products against its AI infrastructure requirements. Initial suppliers include Tesla and Vertiv, but qualification does not replace engineering for an individual site.
NVIDIA has introduced DSX Ready to connect its AI data-centre reference designs with specific infrastructure products. The programme begins with battery energy storage systems and cooling distribution units, which support liquid-cooled computing equipment.
In the announcement by Vishal Ganeriwala, NVIDIA lists qualified battery solutions from Hitachi Energy, LG Energy Solution and Tesla. Its initial cooling suppliers are LG Electronics, LiquidStack and Vertiv.
What the qualification covers
Battery suppliers run required tests and submit supporting data for NVIDIA’s review and approval within a defined qualification boundary. Cooling suppliers use a self-qualification test suite to check whether a particular offering meets NVIDIA’s applicable functional requirements.
The distinction matters for builders choosing equipment. A qualified product has met the relevant category requirements; it has not been cleared for every possible facility. NVIDIA explicitly says battery qualification neither replaces site-level engineering nor establishes site-level stability.
NVIDIA says the programme should reduce integration risk and make products easier to evaluate. Builders still need to assess their own configuration, operating needs and physical constraints. The company plans to add further infrastructure and software categories over time.
Teams building around NVIDIA DSX can use the qualification to narrow equipment choices against a common design. They must still check how batteries and cooling fit their facility, rather than treat the label as approval of the whole installation.
What to watch next
- Creators can bring draft review, presentation choices and audience testing into the same channel-management app. The planned opening-cut tests would let them compare actual edits, rather than judge a video solely by its title and thumbnail.
- An assistant that remembers across devices needs somewhere to keep personal context. This design would move that storage into the cloud while leaving access dependent on users’ devices, rather than relying only on the provider’s account controls.
- Teams building around NVIDIA DSX can use the qualification to narrow equipment choices against a common design. They must still check how batteries and cooling fit their facility, rather than treat the label as approval of the whole installation.
The takeaway
Creators can bring draft review, presentation choices and audience testing into the same channel-management app. The planned opening-cut tests would let them compare actual edits, rather than judge a video solely by its title and thumbnail.
The editor’s view
An assistant that remembers across devices needs somewhere to keep personal context. This design would move that storage into the cloud while leaving access dependent on users’ devices, rather than relying only on the provider’s account controls.
Sources & further reading
- Ivan Mehta, TechCrunch: YouTube releases new AI features for creators within its Studio app ↗
- Google Private AI Compute Team: Advancing Private AI Compute with secure, server-side memory ↗
- Vishal Ganeriwala, NVIDIA Blog: NVIDIA Launches DSX Ready to Qualify Power and Cooling Products for AI Factories ↗