Redefine Your Information Access- Leveraging AI for Smarter Subscriptions

Introduction

In this digital age, we have a plethora of information sources:

• Multimedia: News websites, Medium, e-books, research reports, online journals
• Social Media: Twitter, Instagram, Facebook, LinkedIn
• Streaming Media: YouTube, TikTok, podcasts
• Forums: Reddit, Quora
• Instant Messaging Tools: WhatsApp, Telegram, WeChat

Which source do you find most valuable?

For me, any source that allows me to actively subscribe, filter, and search is valuable. However, many sources, driven by commercial interests, insert personalized ads or dopamine-stimulating content, diminishing the effect of active selection.

The good news is that RSS can effectively mitigate these issues and has gained new vitality in the era of large language models (LLMs).

Traditional Information Retrieval Process and Its Issues

RSS is an information aggregation method free from ads and algorithms, allowing subscription and reading in any product that supports RSS. Note that ad-free and algorithm-free are not absolute, depending on the purity of the source provider. Here’s an excellent introduction to RSS.

Before the LLM boom, my information retrieval process was: subscribe to sources => skim through => save for later reading => read the full text.

For RSS subscription and skimming, I used: RSSHub + RSSHub Radar + FreshRSS + Fluent Reader + Unread.

FreshRSS managed and stored subscription data, exposing the Fever API for readers (Fluent Reader, Unread) to subscribe with one click. RSSHub found RSS sources, and Radar detected website RSS feeds. If no source was found, I tried various “anything to RSS” services, as outlined here.

For later reading, I used Pocket.

I used this setup for over a year but faced several issues:

• Inability to subscribe to all interested sources. Some sites couldn’t be found or converted to RSS.
• Excessive noise. Interest in a source doesn’t mean interest in all its content, and some content isn’t needed at that moment.
• Low reading efficiency:
• - Many articles don’t need to be fully read; a summary is often enough.
• - Foreign language sources are hard to read efficiently.
• Inconvenient search in later reading tools. While called “later reading,” the right time might be when solving a specific problem, making it hard to search with imagined keywords.

Changes Brought by AI

The issues mentioned improved with the development of LLMs.

Tools like lightfeed made converting anything to RSS easier and provided custom prompt filtering to reduce noise.

Reading efficiency also improved with AI summarization tools like Readwise Reader.

Newer later reading tools like Cubox and iki.ai automatically tag content, making search easier.

Additionally, structured reading tools like elmo and zKnown can output mind maps and key points, enabling conversational reading.

The information retrieval process transformed into: subscribe to sources => organize information => skim through => save for later reading => structured + conversational reading => read the full text. Each step optimized for efficiency.

However, there are still some issues:

• High costs. The Readwise Reader subscription averages $10 per month, and Lightfeed costs $3 per month for 5 sites.

• Dispersed tools. Switching between Lightfeed, Readwise, elmo, etc.
• Insufficient filtering. Lacking keyword filtering, which is often enough without wasting AI calls.
• Limited summary customization. While Readwise Reader supports global custom prompts, it can’t customize for specific information types.
• No title translation.

Most importantly, there’s no sense of order:

• Accumulating unread information over time creates a psychological burden, similar to a to-do list with hundreds of overdue tasks that can’t be cleared with one click but must be processed individually.
• As subscriptions grow, reading during fragmented time feels like being overwhelmed by an information flood, increasing anxiety. It’s like an endless social media feed.

To address these issues, I developed Tidyread.

A New Solution

Tidyread introduces a new concept: Recipe. Under a Recipe, you can subscribe to multiple sources, such as AI-related ones, and customize summary prompts. Information is pushed in a digest format at specified periods.

This reduces the concept of individual information entries, replacing countless red dots with a few unread digests daily. If you don’t want to read for a while, you can turn it off with one click or disable a specific Recipe.

We are developing a filter feature that supports keyword and prompt filtering, allowing specific prompt selection for filtered content. This will be available soon.

If you face:

• Information anxiety
• Information overload
• Inefficient reading

If you want to:

• Establish a sense of order in information reading
• Improve reading efficiency
• Share and discuss information sources with other high-quality users

Join our global beta test. Participate by visiting our website, clicking the Join Waitlist button, and filling out the form (about 2 minutes). We will notify you upon submission and promptly send beta invitations and tokens.

We can’t guarantee the product will fully meet your needs now, but our goal is to serve those facing these issues and looking to create customized information digests. We look forward to your questions and suggestions to make Tidyread even better 🥰.

To interact with the core team or other users with a passion for information retrieval, join our Discord or contact us anytime via email: tidytiny.ai@gmail.com.

Appendix

Want to know the product story of Tidyread? Check out this article.