NextFin News - Acti is betting that the most useful AI on a smartphone will not live in another app at all. Instead, the Singapore-based startup has launched an agentic keyboard for iOS and Android that can do more than finish sentences: it can surface information, trigger actions, and carry AI into the apps people already use for email, messaging, and social media. The core idea is simple. If AI is strongest at understanding intent, then the fastest place to use it may be the place where intent first appears.
That pitch matters because it aims at one of mobile computing’s oldest habits: typing. Most consumer AI tools still ask users to leave the app they are in, open a chatbot, ask a question, and then return to the original conversation or workflow. Acti is trying to remove that detour. By embedding AI inside the keyboard, the company wants to make assistance feel like part of the act of typing rather than a separate destination.
Young Wang, Acti’s founder and chief executive, argues that the problem is fragmentation. He says users constantly switch between apps just to get AI help, which leaves context scattered across separate services. Acti’s answer is a keyboard that sits across those apps and builds a context layer for the user rather than the platform. That is a powerful framing, because it shifts the value proposition from “another AI app” to “the interface that already sits everywhere.”
The launch also leans on Wang’s background in keyboards, which is unusually relevant. He spent a decade at Baidu, where he helped grow Facemoji Keyboard to more than 300 million daily active users. In a category where usage depends on habit and distribution, that history matters. Keyboard software is not a place where people casually experiment. It is a place they use repeatedly, often without thinking, and that makes it a tempting entry point for AI that wants to feel invisible.
Acti says the product can do practical things in-line. A user can ask for a nearby restaurant recommendation without leaving a conversation, or pull up a stock price while discussing a company in chat. The app also includes preset Skills, including one that translates a message by long-pressing the letter T and another that fires off a meeting link with the letter C. Users can create their own Skills in plain language, which turns the keyboard into a shortcut system for repeat tasks.
The company is trying to make that feel safe by keeping context local-first. Acti says personal context stays on the device by default and that it does not access or store private messages or conversations unless a feature requires external processing and the user explicitly invokes it. That is a crucial promise, because a keyboard is one of the most sensitive places a company can insert AI. It sits close to chats, payments, work correspondence, and the small details people type before they even decide they want help.
Under the hood, Acti says it uses Google’s Gemini models. The company describes that choice as a balance of intelligence, speed, reliability, multilingual performance, and cost efficiency. It also says the app will eventually generate revenue through subscriptions that unlock more advanced models, higher daily usage limits, and other premium features. That combination suggests the startup is not trying to sell AI as a standalone novelty. It is trying to sell frequency: a recurring reason to keep using the keyboard because it saves time across everyday tasks.
Why The Keyboard May Be A Stronger AI Surface Than A Chat Window
Acti’s thesis is that AI becomes more useful when it appears at the moment of intent, not after a user decides to leave the app and search for help elsewhere. That is not just a design preference. It is a product strategy built around compression. Every time a user copies text, switches apps, or recreates context in a chatbot, the workflow loses time and momentum. A keyboard that can act directly in the flow of typing has the potential to remove several steps at once.
The company’s examples make the argument concrete. If a friend asks where to eat nearby, the keyboard can surface a recommendation immediately. If someone mentions a stock, the keyboard can bring up the live price in the conversation itself. Those are small actions individually, but they point to a larger thesis: the keyboard is not merely a text box. It is a control layer for the language that already moves through a phone.
That is also why the local-first design matters so much. A keyboard that can see across apps is only valuable if users trust it enough to let it see across apps. Acti says private context stays on device by default unless the user explicitly invokes a feature that requires external processing. The company is making a familiar privacy trade-off, but in a form factor where the stakes are unusually high. A keyboard touches the most personal parts of a phone, and people will not tolerate vague assurances for long.
The startup’s pitch also benefits from a simple behavioral fact: a keyboard does not need to create a new habit. It inherits an old one. A separate AI assistant has to be summoned. A keyboard extension can sit there every time the user types. That makes the product easier to reach, even if it does not guarantee that users will keep returning. In mobile software, distribution is often less about downloads than about repetition.
“Today’s AI agents are fundamentally limited because user context stays fragmented across separate apps,” Young Wang said.
Acti “sits across all of them, which is why we can build a context layer that genuinely belongs to the user instead of the platform,” Wang said.
The important question is whether that interface can remain broadly useful once the novelty wears off. If the keyboard only handles a handful of obvious tasks, it risks becoming a clever utility. If it becomes the default place to translate, summarize, search, and trigger actions across apps, it starts to look like infrastructure. That distinction will decide whether Acti is a feature or a platform.
Privacy, Models, And Monetization Will Decide Whether The Idea Scales
Acti’s biggest hurdle is not explaining the concept. It is proving that users will let an AI keyboard sit close enough to their digital lives to be useful without becoming unsettling. That is why the company’s local-first claim is so central. The product’s value depends on access to context, but its adoption depends on restraint.
The keyboard is especially sensitive because it lives at the boundary of so many high-value mobile moments. People use it for private chats, work messages, shopping, and the half-finished thoughts they would rather not expose unnecessarily. Acti says it does not access or store private messages or conversations unless the user explicitly calls a feature that requires external processing. That is the right kind of promise for this category, but it will ultimately be judged by real usage, not by launch-day language.
Acti’s reliance on Google’s Gemini models also points to the operational demands of the product. A keyboard has to feel fast, consistent, multilingual, and cheap to run. Any latency is visible immediately. Any quality drop is felt in the middle of a sentence. The model layer therefore has to be good enough to disappear into the typing experience, which is a much harsher test than the one most chat apps face.
The business model is equally important. Acti says it plans to earn revenue through subscriptions that unlock more advanced models, higher daily usage limits, and premium features. That is a familiar consumer software model, but it only works if the keyboard becomes noticeably better than a standard keyboard every day. Users will not pay for a novelty. They will pay for a repeated time-saving advantage.
Wang’s background suggests he understands that dynamic better than most founders would. He said he spent roughly a decade at Baidu building Facemoji Keyboard to more than 300 million daily active users. That is the kind of experience that teaches a founder how hard it is to turn a keyboard into habit, and how important it is to make each interaction feel immediate and natural rather than promotional.
Wang said he spent a decade at Baidu growing Facemoji Keyboard to over 300 million daily active users.
If Acti can combine that distribution instinct with a privacy posture that users trust, the company has a real shot at making the keyboard a durable AI surface. If it cannot, the product may still be clever, but clever is not enough in a category that depends on daily use. The keyboard has to become indispensable, not just interesting.
What Acti Signals About The Next Phase Of Mobile AI
Acti’s launch fits a broader shift in consumer AI. The first wave was dominated by chat interfaces. The next wave is increasingly about embedding intelligence directly into the surfaces people already use. On a phone, that means the most valuable AI products may be the ones that disappear into the workflow instead of asking users to leave it.
The competitive landscape is still open. AI could become more tightly tied to operating systems, messaging apps, browsers, or keyboards. Acti is arguing that the keyboard has a special advantage because it is universal, constant, and close to the moment of intent. That does not make it the only winning surface, but it does make it a serious one. The interface with the least friction often has the best chance of becoming the default.
For the broader market, the significance is that AI startups are moving from standalone experiences toward infrastructural ones. The question is shifting from whether a model can answer a prompt to whether the product can live inside the place where the user is already working. Acti is trying to prove that mobile typing is one of those places.
The company has not disclosed public traction, and its launch is still early. But the strategic bet is clear. If AI agents are going to be used continuously on phones, they need an interface that users touch continuously. The smartphone keyboard is one of the few candidates that fits that requirement.
That makes Acti’s launch less about one startup and more about a likely direction of travel. The next important AI surface on a phone may not be a separate app at all. It may be the keyboard people already open before they think to ask for help.
Explore more exclusive insights at nextfin.ai.
