Singapore-based Acti launched an agentic keyboard for iOS and Android on Tuesday and disclosed $5.3 million in seed funding led by BITKRAFT Ventures. The app, powered by Google's Gemini models, embeds AI agents directly into the system keyboard so users can trigger actions inside any app — email, messaging, social — without switching to a separate chatbot. Founder and CEO Young Wang previously grew Baidu's Facemoji Keyboard to 300 million daily active users, and he is betting the keyboard is the right surface for consumer AI.
The pitch is that fragmentation is the core constraint on today's AI agents. A user typing in a messaging app has to leave the conversation, open a chatbot, ask the question, copy the answer, and come back. Acti sits in the input layer that crosses every app, which the company argues gives it a structural claim on user context that platform-bound assistants don't have.
The product centers on a feature called Skills — programmable shortcuts that map a single key to a multi-step task. Long-pressing "T" translates a message; "C" generates a meeting link. Users describe what they want in plain language and Acti builds the Skill, no code required. Early access testers produced more than 1,000 Skills in less than two weeks before launch, including shortcuts for real-time World Cup data and Polymarket links.
Key facts
- 01Singapore-based Acti launched an agentic keyboard for iOS and Android on Tuesday, powered by Google's Gemini models.
- 02The company closed a $5.3 million seed round led by BITKRAFT Ventures.
- 03Early access testers built more than 1,000 Skills — programmable keyboard shortcuts — in less than two weeks before launch.
- 04Founder Young Wang previously grew Baidu's Facemoji Keyboard to 300 million daily active users.
- 05CTO Mike Sun was the founding technical lead behind Baidu's Yike Album, which scaled to 10 million daily active users.
Skills can stay private or be published to a marketplace, which Acti positions as a future monetization layer alongside subscriptions for higher usage limits and more advanced models. Wang said Gemini was chosen for its balance of intelligence, speed, multilingual coverage, and cost — the last point matters when every keystroke could trigger an inference call.
Wang frames the bet as a generational rebuild of a foundational interface. He spent roughly a decade at Baidu before starting Acti, and Facemoji's 300 million daily active users gave him an unusually direct read on what people actually type and why.
Privacy is built around a local-first model. Acti says personal context stays on-device by default, and the app does not access or store private messages, conversations, or personal context unless the user explicitly invokes a feature that needs external processing. That positioning matters for a product that, by design, sits in front of every text input on the phone — a surface that has historically attracted regulatory attention when handled carelessly.
BITKRAFT Ventures led the $5.3 million round. The Acti team includes CTO Mike Sun, the founding technical lead behind Yike Album, Baidu's cloud-photo platform that scaled to 10 million daily active users, and CSO Junbo Yang, who joined from HashKey Capital after leading dozens of consumer investments.
“We backed Acti because this team has a real shot at owning the next phase of human-computer interaction.”— Jonathan Huang, Partner at BITKRAFT Ventures
The risks are the usual ones for an AI-layer startup pinned to someone else's model. Acti's economics depend on Gemini pricing holding or falling — a reasonable bet given the broader trajectory of inference costs, but not guaranteed. The keyboard category is also brutal: iOS and Android both make third-party keyboard adoption friction-heavy, and users are slow to switch off the defaults. Acti has to convert keyboard installs into agent usage frequent enough to support a subscription.
The more interesting question is whether the keyboard becomes a serious AI distribution channel at all. The consensus bet across the industry has been on assistants — ChatGPT, Gemini app, Claude — as the front door. Acti is wagering the front door is wherever the cursor is already blinking, and that owning the input layer beats owning the chat layer because intent shows up in text before it shows up in a query. If that thesis holds, the next round of consumer AI competition runs through IMEs and input surfaces rather than chat windows, and the incumbents with the deepest model access — Google on Android, Apple on iOS — have an obvious answer waiting in the wings.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
