
With the launch of the Galaxy Z Fold 8, Samsung is also shipping its first phone with Gemini Intelligence. I managed to get it working easily, but I don’t understand why it would work for most people.
Subscribe to 9to5Google on YouTube for more video content.
After Pre-IO was introduced at the Android Show, there was little information from Google regarding Gemini Intelligence. It was expected to debut with the upcoming Pixel 11, but it will probably launch alongside the Korean tech giant’s latest foldable devices and smartwatches, as part of a close partnership with Samsung.
We were able to spend some time with the foldable device before its launch, but were told that Gemini Intelligence would not be available until after the device’s launch. However, this may have been an oversight, as Samsung phones come with certain services pre-installed, so the agent AI tools could be used to perform some basic operations.
It comes with a suite of apps, including Expedia, so we were able to recreate the demo shown off before Unpacked. It’s super simple and can automate tasks like travel, flights, hotels, and even booking Ubers if you have a rideshare app installed. Without risking hundreds of angry Uber drivers trying to take me on an impossible trip, I stuck with a simple demo of booking a flight from an airport to a specific destination, and to my surprise, it worked.
This is effectively an expansion of the app automation feature, which was released at the same time as the Galaxy S26 and was limited to a few apps primarily aimed at food ordering, but now works with over 40 apps and is integrated into Gemini Intelligence. This new enhancement allows us to better understand context and intent. App Automation was and still is if-then-else based and limited to selected apps. In theory, Gemini Intelligence should work with any app, but it’s not available everywhere yet.


In case you’re wondering, the process of launching Gemini Intelligence is indistinguishable from regular old Gemini. Yeah, it felt weird typing that.
Gemini will appear when you press and hold the power button or say a named wake phrase. From here, all you have to do is request a specific task. So in my case, I simply asked the AI to book me a flight from London Heathrow to Dublin a week in advance, similar to the demo Samsung showed me before I could try out the new hardware.
Good news. It works.
However, I find that this needs to be very accurate. If I say, “Book a flight from location X to destination Y,” Gemini will show me a text feed with prices and information about upcoming flights, or provide information about the airline’s services.
Adding a day of the week or date modifier indicates when Gemini Intelligence “activates” or wakes up. That being said, there are cases where it works without this additional snippet of information, which confused me even more. Technically, it’s still in beta, so out of the 20-plus times I tried it, Gemini occasionally asked me which app I wanted to use to book flights and hotels. Further exploration was aborted because the app could not be installed on the device.
One thing that I found a little strange is that you don’t have to confirm anything before you start taking control. As long as you have an app installed on your phone that can do what you want or need, it will do the job. Similar to App Automations, Gemini simply displays a “Working on Task” notification next to the app you’re using. big[進行状況を表示]You can see this automation unfold by tapping the button. More details will be explained later.

What was immediately interesting was that on the Samsung phone, the pop-up looked nothing like the preview we saw on the Android show. Popups look like regular prompts and notifications that fit in with the rest of One UI. The company thought it would put a unique visual twist on this. Rather, it looks, well, rudimentary.
The only obvious change is the colorful glow emanating from these pop-ups and notifications. The status bar icon was not displayed while the task was running. I like that it explains step by step what’s going on. For example, you can see what Gemini is typing in the search box or the time of the selected flight.
Google sells this feature automatically, and it works in the background. Again, just like App Automations, you can always step in and take over if needed. You can also monitor the entire process within the Gemini app, as I’ve done several times. Get a complete explanation of what Gemini is doing on your app screen running in siled windows.
My biggest issue with how that works is how long it takes. In some cases, it may take several minutes of preparatory work before you are asked to intervene. My phone can probably do this faster and more accurately than Gemini. I’m scratching my head as to who this really benefits.
For example, I couldn’t enter anything about the exact flight time or even select my preferred airline. Gemini Intelligence never asked me for these details before activation, nor did it ask me during the process. Sites like Expedia can offer thousands of flights from different airlines, so this feels like a significant omission. Humans don’t make that mistake because these parameters are obviously very important to humans.
Like I said, you can always jump in and course correct. Once that’s done, let Gemini take over again to complete the task.
I’m not sure if this applies to all Gemini Intelligence automations, but it should take over once you get to the customer data screen, such as adding passenger information. This may not be true for all tasks, and I don’t know if you’re logged in to the app, but if you are, it’s probably filled in automatically.
Does it really save time?

I realize that I have only used a small portion of the potential of Gemini intelligence. Google said it could do more, like take a photo of a brochure and have it search for it on your device, but instead you had to try to do it using voice commands. I was also unable to receive emails for PC parts and add them to cart on PCpartpicker.
Even if that were the case, I’m not sure this would be of much value to all but the most terminal online tech geeks. AI is great when directed correctly, but there is still too much manual work involved. This feels like a perfect example of that.
Gemini’s intellect is clearly at work, and if you give it clear guidance, it will work quite well. Still, I don’t feel confident that I can handle important tasks like booking a flight or delivering food without going through all the intricacies manually.
That’s a great concept. I don’t think many people use this concept very often. Maybe there are special cases like waiting in line to buy tickets to a concert or checking in to something. Maybe you find yourself booking an Uber, but you’re not saving that much valuable time. I don’t think anyone I know would use or trust Gemini to book a flight based on a brochure or advertisement they found, and I don’t think I would either.
This is like hitting a nut with a sledgehammer because the minimal effort required to perform these tasks is so trivial. Yes, it works, but it’s inefficient and, dare I say it, a bit pointless.
Time to soften my criticism a little. App Automation has been available in the US and South Korea since the S26’s launch, so I didn’t have a chance to try its precursor. It also lacked some of the key integrations and personalization to really leverage the agentic aspects of Gemini Intelligence.
This was a very basic demo and was not truly representative of the “full” experience due to various limitations of pre-release units. Still, I can’t help but feel that Gemini Intelligence is a cool tech demo and nothing more. I’m sure a lot of people won’t agree, but in a way I agree, when you use a full suite for an extended period of time and something “clicks” for you. Again, there needs to be a strong message about how it will benefit the average person who is still enjoying their day-to-day work.
Let us know what you think in the comments section. I’m interested in what people want from agent AI services that run on their phones, and I’d love to hear what you’re planning.
Follow Damien: Threads, Blue Sky, Instagram


FTC: We use automated affiliate links that generate income. more.
