EchoFlow

Product Introduction

1. Definition: EchoFlow is an open-source, native Android AI chat application that operates on a strict Bring-Your-Own-Key (BYOK) model, powered by the OpenRouter API aggregator. It is a client-side interface for interacting with large language models (LLMs).
2. Core Value Proposition: EchoFlow exists to provide a private, customizable, and truly Android-native AI chat experience. Its primary purpose is to eliminate server-side data retention and tracking, giving users full control over their chat history and API key, while delivering a distinctive, expressive UI that diverges from generic web-based chat app designs.

Main Features

1. Material 3 Expressive Design: The app is built entirely on Google's Material 3 Expressive design system, not just the base Material You. This implementation features custom "scalloped" UI shapes, physics-based spring animations for taps and transitions, bold typography, and soft, generative surfaces. How it works: The app uses Android's dynamic color theming (Material You) on Android 12+ to generate a palette from the user's wallpaper, which is then applied across all components within the Material 3 Expressive framework for a cohesive, playful, and non-rectilinear look.
2. 100% Local Chat Storage & Offline Access: All conversation histories are stored exclusively on the user's Android device. How it works: The app uses a local database (like SQLite or Room) on the device to save every chat thread. This data is never uploaded to any server, enabling complete offline access to past conversations. The BYOK API key is also stored locally in the app's secure storage.
3. OpenRouter-Powered BYOK Multi-Model Support: EchoFlow functions as a universal interface for any model available via OpenRouter. How it works: Users paste their OpenRouter API key (sk-or-v1-*) once. They can then add any model by its specific ID (e.g., openai/gpt-4, anthropic/claude-3.5-sonnet, meta-llama/llama-3-8b-instruct). The app can search, add, and switch between these models mid-conversation via a bottom sheet interface. Free models work directly without additional keys.
4. Real-Time Streaming & Reasoning Traces: Replies are displayed token-by-token as they are generated. How it works: The app establishes a streaming connection with the OpenRouter API (using server-sent events). It renders the response in real-time with a live caret. For models that support it (like Claude 3.5 Sonnet), a collapsible "thinking" panel can display the model's reasoning trace as it processes the query.
5. Rich Markdown and Attachment Support: The chat interface fully renders complex content. How it works: The app includes a markdown renderer that properly formats code blocks (with syntax highlighting), tables, lists, bold/italic text, and links within chat bubbles. It also supports sending image attachments (photos/screenshots) from the device's gallery to vision-capable multimodal models via the API.
6. Seven Distinct App Themes & Appearance Modes: Beyond Material You, the app offers seven curated color palettes. How it works: Users can select from predefined palettes (Ocean, Forest, Sunset, etc.) in both light and dark modes. These themes override the system dynamic color if chosen, applying a consistent accent color throughout the entire application UI.

Problems Solved

1. Pain Point: Privacy concerns and data sovereignty. Most commercial AI chat apps (e.g., ChatGPT, Google Gemini) store conversations on their servers, raising issues of data privacy, security, and potential tracking. EchoFlow solves this with its local-only storage and zero telemetry guarantee.
2. Pain Point: Generic, repetitive UI design across AI apps. Many apps use a standard shadcn-style web UI, leading to a homogenous user experience. EchoFlow addresses this with its Material 3 Expressive design, offering a unique, playful, and deeply Android-native aesthetic.
3. Pain Point: Model lock-in. Users are often confined to a single provider's models within an app. EchoFlow solves this by being a BYOK OpenRouter client, allowing users to access multiple LLM providers from a single interface with one key.
4. Pain Point: Lack of offline functionality for chat history. EchoFlow ensures full offline access to all past conversations stored on-device.
5. Target Audience: Privacy-conscious individuals, Android power users, AI developers and researchers testing multiple models, users dissatisfied with mainstream AI chat app designs, and individuals in regions with restrictive internet who benefit from offline history access.
6. Use Cases: Conducting sensitive or private conversations without server-side logging; testing and comparing responses from different frontier and open-source models (GPT-4, Claude, Llama) side-by-side; accessing AI chat capabilities with a simple, one-time key setup; using a visually distinct and customizable chat interface; reviewing previous AI conversations during commutes or offline.

Unique Advantages

1. Differentiation: Compared to mainstream apps like ChatGPT or Google Gemini, EchoFlow has no account requirement, no usage tracking, and no server-side chat history. Unlike other open-source clients, it is not a web wrapper but a fully native Android application built specifically for the platform, leveraging its native UI toolkit and design philosophy (Material 3 Expressive) for optimal performance and feel.
2. Key Innovation: The core innovation is its privacy-first, BYOK architecture combined with a first-party, expressive Material 3 Android implementation. It decouples the AI model (accessed via OpenRouter) from the data layer (100% local), while wrapping the interaction in a UI that is both highly customizable and fundamentally "Android-native" in its motion and aesthetics.

Frequently Asked Questions (FAQ)

1. What is BYOK and how does EchoFlow use it? BYOK stands for "Bring Your Own Key." EchoFlow requires you to obtain an API key from OpenRouter (a service that provides access to hundreds of AI models). You paste this key into the app once. EchoFlow then uses your personal key to make API calls directly from your device to OpenRouter, meaning you pay the API provider directly and no third-party (including EchoFlow) handles your conversations or key after setup.
2. What AI models can I use with EchoFlow? You can use any model available on the OpenRouter platform. This includes popular models from OpenAI (GPT-4o, GPT-4), Anthropic (Claude 3.5 Sonnet), Meta (Llama 3), Mistral, Google (Gemini), and many open-source models. You add them by their model ID within the app.
3. Is EchoFlow free to use? The EchoFlow app itself is free and open-source. However, using it incurs costs based on the OpenRouter API pricing for the specific models you query. Some models on OpenRouter are free to use, while others are pay-per-token. EchoFlow displays which models are free.
4. How is my data kept private? EchoFlow is designed with a "Private by Default" principle. Your API key and all chat histories are stored only in the app's local storage on your device. The app contains no analytics, no telemetry, and has no server component. Your data never leaves your phone.
5. Does EchoFlow work offline? While the AI chat functionality requires an internet connection to communicate with the OpenRouter API, all your past conversations are stored locally and can be accessed and read offline at any time.