Your Keyboard Just Became the Most Important AI Interface
I’ve used a smartphone keyboard for fifteen years. Never once did I think of it as an interface for intelligence.
Until now.
Acti didn’t just update a keyboard. They rewired the most intimate, frequent, and unconscious interaction we have with our phones—and turned it into the first real agent surface. No app switch. No chat window. No asking. Just… action.
I’ve watched AI apps come and go. ChatGPT, Claude, Gemini—they all ask me to leave what I’m doing. To open a new thing. To type a prompt. To wait. To wait again.
Acti doesn’t ask. It listens. And then it does.
The Silent Bottleneck
You know this feeling.
Friend texts: "Where’s a good Thai place near me?"
You open Maps. You open Google. You scroll. You copy. You paste. You reply.
Three minutes later, you’re back in the chat. The moment’s gone. The vibe’s broken.
That’s the bottleneck. Not the AI. Not the model. It’s the friction between where your attention is and where the AI lives.
Acti’s CEO, Young Wang, says it plainly: "Today’s AI agents are fundamentally limited because user context stays fragmented across separate apps."
He’s right.
We’ve spent the last three years building AI as a separate room. Acti’s realizing it should’ve been the doorway.
Skills: No Code, Just Intent
Here’s the part that cracked me open.
You don’t need to learn Python. You don’t need to prompt engineer. You don’t even need to know what an LLM is.
You long-press "T"—and your message auto-translates to Spanish.
You long-press "C"—and it drops a Zoom link into your DM.
That’s not a feature. That’s a new muscle.
Early testers built over 1,000 of these—called "Skills"—in two weeks. One person made a Skill that pulls live World Cup stats into a chat. Another built one that links to Polymarket bets. No code. Just: "When I type 'weather' in a text, show me the next 24 hours in my friend's city."
And Acti builds it.
This isn’t automation. It’s translation. Text becomes intent. Intent becomes action. And the keyboard? It’s the translator.
Local-First, Not Just Privacy-First
You’re probably thinking: "Wait, this thing sees all my messages?"
It doesn’t.
Acti’s architecture is local-first. Your private DMs? Your work emails? Your late-night rants about your boss? They stay on your device. Period.
The AI only wakes up when you trigger a Skill. Only then does it send a sanitized, intent-only snippet to the cloud. Nothing else. No memory. No log. No shadow.
That’s not a marketing line. That’s the difference between a tool and a spy.
Wang’s team didn’t just build an AI keyboard. They built one that respects the fact that your phone isn’t a data farm.
The Team That Built This
This isn’t a random startup.
Young Wang spent a decade at Baidu, growing Facemoji Keyboard to 300 million daily users. That’s not a product. That’s a cultural artifact. He didn’t just design a keyboard—he studied how people think while typing.
CTO Mike Sun built Baidu’s Yike Album, which scaled to 10 million daily users. CSO Junbo Yang led dozens of consumer investments at HashKey. This team doesn’t just know tech—they know how people use it.
And they’re not chasing hype. They’re solving a real, daily annoyance.
$5.3 Million and a New Paradigm
BITKRAFT Ventures just put $5.3 million into Acti. Their partner, Jonathan Huang, said it best: "We backed Acti because this team has a real shot at owning the next phase of human-computer interaction."
That’s not hype. That’s a declaration.
We’ve been sold AI as chatbots. As copilots. As assistants you summon.
Acti says: No. It’s the thing you never notice—until it works.
The Future Is Invisible
The best interfaces don’t announce themselves.
You don’t think about the steering wheel when you turn left.
You don’t think about the light switch when you flip it.
Acti’s keyboard doesn’t ask you to think. It just… does.
The next time someone texts you a stock name, and you see the price drop into your chat without lifting a finger?
That’s not magic.
That’s the keyboard becoming the interface.
And it’s already here.
I’m not waiting for the next AI app.
I’m waiting to see what Skill I’ll build next.
The Tech Behind the Magic
Under the hood, Acti runs on Google’s Gemini models—not because they’re the flashiest, but because they’re fast, reliable, and multilingual enough to handle the global messiness of real conversation. Wang told TechCrunch they chose Gemini for its balance of intelligence and cost-efficiency. No need to overpay for a model that hallucinates your friend’s Thai restaurant into a sushi bar.
The Skills engine? It’s the real breakthrough. You don’t write code. You don’t even write prompts. You just say what you want in plain language: "When I type 'weather' in a text, show me the next 24 hours in my friend’s city." Acti’s system parses that, maps it to available APIs, and builds a trigger. No backend. No API keys. Just intent.
And here’s the kicker: early testers built over 1,000 Skills in two weeks. That’s not a feature. That’s a movement. People aren’t just using the keyboard—they’re reprogramming it. One user built a Skill that auto-responds to "I’m running late" with a Google Maps ETA and a GIF of a turtle. Another made one that pulls Polymarket odds into a group chat before a football game. These aren’t power users. These are people who just hate switching apps.
The Privacy Lie We’ve Been Told
Every AI app says "privacy-first." Most mean "we won’t sell your data." Acti means something else: "we won’t touch it unless you ask."
Your DMs? Your Slack threads? Your text to your mom about your dad’s surgery? All stay on-device. Full stop. The AI only gets a stripped-down version of your intent—"translate this," "find a Thai place," "send this link"—and nothing else. No history. No profile. No behavioral tracking.
That’s not a feature. It’s a rebellion.
We’ve been conditioned to think AI needs to know everything about us to work. Acti proves that’s a lie. You don’t need to hand over your entire life to get something useful. You just need to be clear about what you want.
Why This Matters More Than the Next Chatbot
We’re drowning in AI assistants. But none of them live where we live.
You don’t open ChatGPT to reply to a text. You don’t summon Claude to book a dinner reservation. You don’t ask Gemini to translate your cousin’s message from Tagalog.
You open your keyboard.
And now, for the first time, that keyboard has a brain.
This isn’t incremental. It’s foundational. The keyboard is the last universal interface on mobile. Every app has a button. Every website has a menu. But the keyboard? It’s always there. Always open. Always listening.
Acti didn’t build an AI app. They built the first AI-native input layer.
And once that layer exists, everything else becomes a plugin.
The Next Step: Your Skills, Your Rules
The Skills marketplace is coming. You’ll be able to share yours. Download others’. Maybe even sell them.
But here’s what I care about: the next time you’re in a group chat and someone says, "Who’s got the WiFi password?"—and your keyboard just drops it in without you saying a word.
That’s not convenience.
That’s the end of friction.
And I’m not just ready for it.
I’m already using it.
