iOS and Android

An AI chatbot that runs entirely on your phone.

Silsila AI is an on-device AI chatbot. It runs a language model directly on your iPhone or Android phone, with no account and no cloud. Once the model is loaded, the chat works with no network connection, so your conversations are processed on the device itself.

Runs fully offline once the model loads. Your conversations stay on the device.

Silsila AI welcome screen with a Start a New Chat button on a dark background

100%

on-device chat

0

cloud calls per reply

2

app stores

On-device chat

A chatbot that runs on your phone, not the cloud.

Open the app and start a chat. The language model runs locally on your device, so there is no account to create and no server in the loop.

Silsila AI welcome screen with a Start a New Chat button on a dark background

Real answers, locally

Ask anything. The reply comes from your device.

Silsila AI responds from a model running on your phone: information, brainstorming, drafting, and more, all handled on-device.

A Silsila AI conversation, the assistant replying to a greeting

Loaded into memory

The model lives on the device.

On launch, Silsila AI loads a quantized language model into memory. After that, the chat works with no network connection at all.

Silsila AI loading the on-device model on launch

Simple by design

One screen. Type and go.

A plain chat surface with nothing to configure. Type a prompt, read the response. No menus to learn first.

Silsila AI prompt screen with a write your prompt here input

Stays on your phone

Conversations are processed on the device.

Because the model runs locally, your chats are not sent to a server to be answered. See the privacy policy for the analytics and diagnostics the app records.

Silsila AI settings screen crediting Sayed Ali Alkamel as the developer

Under the hood

A language model, running on the phone.

Silsila AI is native on each platform and runs a Gemma model on-device through Google's MediaPipe LLM Inference. The model is quantized to fit in phone memory, loaded once, then used with no network connection to answer you.

Platforms
Native iOS (Swift) and Android (Kotlin)
Inference
Google MediaPipe LLM Inference
Model
A Gemma model, quantized for on-device use
Network
None to answer a chat once the model is loaded

Versions

Release history

  1. 1.0latestMar 20, 2025

    Initial release: an offline, on-device AI chatbot.

The chat runs on your device. Analytics and diagnostics are handled as the policy describes.

Read the privacy policy