We should all disfavor booking a table at a restauranton memoir of it’s over again the plan back tech corporations are making an strive to solve with the flexibility of man made intelligence. Honor has taken the wraps off of Honor UI Agent—a “GUI-based fully cell AI agent” that claims to deal with initiatives for your behalf by conception the display disguise disguise’s graphical user interface. Its predominant demo to blow their like horns this functionality? Having the agent e book a restaurant, naturally, thru OpenTable.
WIRED had an early opportunity to look the demo ahead of the firm’s keynote at Mobile World Congress 2025 in Barcelona, the place Honor additionally launched its $10 billion Honor Alpha Plot. This prolonged-duration of time idea, envisioned by the Chinese language firm’s fresh CEO Jian Li, is lofty and largely corporate-notify, comprised of targets like “creating an wise cellphone” and “open human potential boundaries and cocreate a new paradigm for civilization.” What it really highlights is Honor’s quick pivot into prioritizing AI development for its suite of personal technology devices.
A GUI Agent
In the demo, an Honor spokesperson asked Honor’s UI Agent to book a table for four people, gave a time, and specified “local food.” (The AI takes location into account and understood that to mean Spanish food here in Barcelona.) What happens next is a little jarring—not in the way Google’s Duplex technology was when it debuted in 2018 and had Google Assistant interact with real humans to make reservations on your behalf.
Instead, you’re forced to stare at Honor’s screen, watching this agent run through the steps of finding a restaurant and booking a table through the OpenTable app. It doesn’t quite feel “smart” at the same time as you wish to look the dreary machinations of the approach at work, though Honor tells me in due course its UI Agent will not need to narrate its homework.
Photograph: Julian chokkattu
It chose a restaurant but then might per chance not total the approach because the distance it chose required a bank card to ascertain a reservation, at which level the user needed to take over. You is seemingly to be versatile for your ask—in a single other example, asking it to e book a “highly rated” restaurant meant it would gaze at opinions with high scores, though the agent doesn’t design to any extent further be taught than that. Or no longer it is no longer scandalous-referencing OpenTable opinions with data from other parts of the gain, in particular since all of this recordsdata is processed on tool and is rarely any longer sent to the cloud.
This roughly agentic man made intelligence is the fresh buzzword in the tech sphere. My colleague Will Knight no longer too prolonged ago tested an AI assistant that can per chance per chance browse the gain and develop initiatives on-line. Google late last yr unveiled its Gemini 2 Ai Model that’s expert to take actions for your behalf. It additionally renews the premise of a generative user interface for smartphones—at MWC 2024, we saw about a corporations working on systems to earn interaction with apps without the utilization of apps at allas a change leaning on AI assistants to generate a user interface as you issued a exclaim.
Honor’s formulation feels considerably like what Rabbit—of the infamous Rabbit R1—is doing with Reveal Modethe place you put together its assistant manually to total a role. There’s no need to get entry to an app’s software program programming interface (API), which is the feeble procedure apps or services and products keep up a correspondence with one but every other. The agent memorizes the approach, permitting you to then field the exclaim and earn it design the job.
Nevertheless Honor says its self-reliant AI execution model is rarely any longer expert to apply strict steps—or no longer it is succesful of multimodal display disguise disguise context recognition to develop initiatives autonomously. As a replacement of having to put together the assistant to be taught each a part of the OpenTable app, it is succesful of conception the semantic aspects of the user interface and need to apply-thru with a multi-step course of to design your request. Honor highlighted that this course of was more economical: “No longer like rivals akin to Apple, Samsung, and Google, which rely on exterior APIs—ensuing in greater operational costs—Honor’s AI Agent independently manages a huge series of initiatives.”
Photograph: Julian chokkattu
While Honor says its UI agent uses in-house execution models, it additionally leverages Google’s Gemini 2 big language model, which is what powers the intent recognition of your exclaim and the “enhanced semantic conception” of what is on the display disguise disguise. Google did no longer allotment any particulars regarding the nature of the collaboration.
Honor says it has additionally partnered with Qualcomm to preserve the details on the tool and kind a deepest recordsdata immoral that learns your preferences over time. The foundation is that at the same time as you are inclined to expose the actual forms of meals in a shipping app, at the same time as you request the agent to expose for your behalf, it’ll spend that context to pick out one thing it is conscious of you adore. The firm says or no longer it is already employing a majority of these AI agents in China.
At its keynote, Honor additionally launched that it’ll convey seven years of instrument updates for its flagship Magic 7 Pro and upcoming devices—matching the software update policies from Google and Samsung for Pixel and Galaxy phones.
It unveiled a handful of new gadgets at the show too, including the Honor Earbuds Open, Honor Watch 5 Ultra smartwatch, Honor Pad V9 tablet, and Honor MagicBook Pro 14 laptop. These devices won’t be sold in the US, like most of Honor’s products, but will be available in other markets. (The brand hosted WIRED at its media event at MWC 2025 and paid for a portion of our reporter’s travel expenses.)
