Acti Launches Local-First Agentic Smartphone Keyboard Powered by Google Gemini Models
Singapore-based startup Acti has released an agentic keyboard for iOS and Android that handles multi-step tasks natively across mobile apps. Powered by Gemini, the keyboard allows users to configure plain-language 'Skills' that trigger automated actions and information retrieval.
Why it matters
By embedding agentic capabilities directly into the smartphone keyboard, Acti removes the friction of switching between multiple apps and chatbots, delivering AI assistance in-context.
TL;DR
- 01Acti is an agentic mobile keyboard for iOS and Android powered by Google Gemini.
- 02Users can create custom 'Skills' in plain language without coding.
- 03A local-first architecture keeps personal data on the device by default.
Key facts
- LLM Backend
- Google Gemini
- Built-in Skills Examples
- T (translate), C (meeting link)
- Target Platforms
- iOS and Android
- Pre-launch Skills Created
- 1,000+
The Plain-Language Skills Engine
Users don't need to know how to code to create a Skill; instead, they can simply describe what they want in plain language, and Acti builds it. For example, assigning a shortcut to a key can execute actions like sharing a meeting calendar link or fetching a stock ticker price. Testers built 1,000+ custom Skills in under two weeks.
Privacy Architecture
Because mobile keyboards record sensitive inputs, Acti utilizes a local-first paradigm. Key strokes, personal chats, and conversation history remain stored on-device. Network requests to Gemini and external APIs are restricted strictly to user-initiated agent skills, mitigating the typical data-scraping concerns associated with third-party keyboards.
✓ When to use
- When looking to consolidate AI workflows directly into text-based mobile communication.