Frontier - Why Develop This Open Source Software
As a typical front-end developer, last year when researching the Rust front-end toolchain and the technical system for LLM application development for the company, I developed a strong interest in these technical fields. Also, due to my past experience as an individual Android mobile application developer, I habitually set a tool-themed product for myself to practice the application development technology of Rust and AI.
The purpose of the product (user value) is to create a more generalized professional software based on actual needs:
- The need for timely intelligence: In this era of information explosion and rapid development in all industries, how can one obtain information in a timely manner in combination with their areas of focus?
- The need for reading efficiency: The writing styles and standards of various articles on the Internet vary greatly. How can an AI narrator express in a consistent style, and how can one not only see the original text but also directly access the refined core viewpoints, conclusions, and logic?
- The need for in-depth understanding and discussion: Reading alone can be lonely. There is a desire for a reading companion feature that allows for discussions with an AI about the content of articles and the exchange of opinions and views.
- The need for no distractions: Eliminate the spam articles brought by advertisements and recommendation algorithms and block the advertisement content in various media articles.
So, in my personal weekend leisure time, I developed a software. The technology stack includes Web and Rust, and the core framework uses Tauri. Solve corresponding problems with appropriate technologies. Tool software should be portable, easy to use, high-performance, energy-efficient, and highly secure.
An extremely fast and lightweight Internet reader client driven by artificial intelligence.
A modern cross-platform application built with Rust, Tauri, and Saga.
I just uploaded this project to GitHub in May ( Github - Saga Reader ). Everyone is welcome to follow and share it.
In the future, I will also have a special topic to record the technical points of this project, and I will gradually share some experiences in large model application development, Rust development, and personal software operation. Everyone is welcome to communicate. Your stars are my great motivation.
What is Qi Rui Think Tank
Qi Rui Think Tank (Saga Reader) is an AI-driven think tank-style reader that automatically retrieves information from the Internet based on user-specified topics and preferred keywords. It uses cloud or local large models for summarization and provides guidance, and it includes an AI-driven interactive reading companion feature that allows you to discuss and exchange ideas about the reading content with the AI.
Qi Rui Think Tank is completely free and open source, which means that all data is safely stored on your own computer and is not controlled by third-party service providers. In addition, you can manage subscription keywords according to your own interests and preferences without being disturbed by advertisements and commercialized content.
Open source address: Github - Saga Reader
Screenshots
The main interface of the reader is simple and elegant, supporting theme switching
Content subscription based on interest themes, allowing for independent and flexible settings to gather information from the global Internet
Collect information in multiple languages, create a full Chinese reading experience, and provide article structure optimization, automatic summarization, suggestions, and reminders
AI reading companion: Have in-depth discussions and exchanges with the personified AI assistant based on the article content
Flexible large model selection: You can choose cloud large models (such as Qinghua Zhipu) and local large models on your personal computer (supporting Ollama)
Technical Implementation
Overall Workflow
- Frontend: Involves multi-component interaction, state management, and internationalization processing. Components share states through Svelte Store, and the internationalization module dynamically switches language resources according to user settings.
- Backend: tauri - plugin - feed - api serves as a Tauri plugin to interact with the frontend and coordinate other business modules. feed_api_rs is responsible for the core business process, linking modules such as llm, recorder, and scrap, and handling complex business logic.
- Cross-module collaboration: After the article is crawled, the scrap module passes the data to the intelligent module, the intelligent module calls the llm for content processing, and the final result is stored by the recorder. The whole process involves the collaborative work of multiple modules.
Core Modules
Each module is both independent and closely collaborative, forming a complex and orderly whole.
intelligent: The article optimization workflow module provides an abstraction of the article optimization process and prompt engineering optimization.scrap: Provides data crawling functions, obtaining information from the Internet by calling mainstream search engines. This module is completely localized by default and does not rely on any third-party services.recorder: Provides local storage functions. User interest prompts, original articles, and post-processing optimized articles are saved in the user's personal computer storage.llm: Provides an abstraction of the internal LLM Provider and adapts to the implementation of various cloud and local large model services.ollama: Runs the local ollama, including functions such as running basic instances, model updates, and management.feed_api_rs: The core capability API and implementation based on the classic facade pattern.tauri-plugin-feed-api: Calls the core capability API of the frontend through tauri commands.types: A module for sharing basic types.
Content Subscription and Update Process
Article Reading and Artificial Intelligence Interaction Process
About Me
I am an Internet veteran who has experienced three waves: the PC Internet, the mobile Internet, and AI applications. I entered the field as an individual mobile application developer in the early days and now work as a professional in the workplace. I have rich experience in product design and R&D. Currently, I am working at Kuaishou, engaged in R&D work in the big front-end system and AI exploration.