Is your inbox
a newsletter landfill?

StackDigest is currently free during the beta period. I'm focusing on getting feedback from early users and refining the product before determining pricing. Beta users will get special consideration if/when I do introduce pricing.

There's currently no hard limit on the number of newsletters you can add. I personally tested it with 200+ subscriptions, and some beta testers have even more. The processing time increases with more newsletters, but the system handles larger collections well.

StackDigest uses the Claude family of models from Anthropic to generate article summaries and categorize content. I chose Claude because it can create concise, accurate summaries in a straightforward writing style. The specific model used can vary based on the task (faster models for categorization, more powerful ones for complex summaries).

No, your content is not used to train AI models when processed by StackDigest. Anthropic has confirmed that they do not use API inputs and outputs to train Claude. This means that when StackDigest sends newsletter content to Claude for summarization, that content is not "copied" or harvested for AI training purposes.

However, be aware that Substack offers no way to turn off or limit RSS feeds for free posts. This means that while StackDigest itself doesn't contribute to AI training, your public RSS feeds can still be scraped by LLMs in the wild. If you want to avoid this broader scraping, your options include paywalling your posts or moving your content to a website where you control RSS access.

Yes! You can adjust how many articles get AI summaries (the default is 7) and set the time period for your digest (daily or weekly). I'm working on more customization options, including the ability to prioritize certain topics or newsletters.

Currently, StackDigest is optimized for Substack newsletters, but testing shows that it can work with some RSS feeds with URLs ending with the extension /feed. Full RSS support for any content source is on the roadmap.

You have three options: add individual newsletter URLs, bulk import a list, or use a bookmarklet to grab your subscriptions from your Substack settings page.

Right now, digests are viewable in the app, you can copy the HTML, or you can email them to your inbox with one click. Multiple export formats are on the roadmap.

Yes. Your newsletter subscriptions and digest history are tied to your account and stored securely. I don't share or sell any user data. The only external service that processes your content is Claude's API for summarization, and as mentioned above, Anthropic doesn't use this data for training.

Absolutely! In fact, there's an excellent tutorial by Peter Yang that shows how to use n8n to extract newsletters from Gmail and create weekly summaries. When I was researching solutions for newsletter overload, several people mentioned they were building their own versions using n8n or similar automation tools.

I chose to build StackDigest as a dedicated tool because I wanted something that could handle 200+ subscriptions efficiently, provide real-time processing updates, and offer features specifically designed for newsletter readers. But if you're comfortable with automation tools and want full control over your digest process, n8n is a great DIY option.

I'd love to hear from you! You can reach me directly through my Substack DMs, email me at karen@stackdigest.io, or by replying to any of my Substack posts about StackDigest. I'm building this in public and actively incorporating user feedback into the development process.

StackDigest is designed specifically to help you keep up with all your Substack newsletter subscriptions. Instead of letting newsletters pile up in your inbox, our AI-powered system automatically collects your subscriptions, identifies the most important articles, and creates personalized digests. You can run daily or weekly digests, use the semantic search to focus on specific topics, and even discover new newsletters using vector similarity across 2,000+ newsletters. It's the perfect solution for staying informed without feeling overwhelmed.

StackDigest provides a centralized dashboard to manage all your Substack subscriptions in one place. You can organize newsletters into collections, assign categories, adjust scoring preferences, and control which newsletters appear in your digests. The browser bookmarklet makes it easy to import all your existing subscriptions directly from your Substack settings page, and you can add new newsletters anytime through URL import or our discovery feature.

StackDigest is perfect if you've turned off email notifications to reduce inbox clutter. We pull articles directly from RSS feeds, so you don't need email notifications enabled. You can generate digests on-demand through our web interface, ensuring you stay updated with your favorite newsletters without any emails. This approach gives you complete control over when and how you consume your newsletter content, without missing important articles.

If you're overwhelmed by too many Substack subscriptions, StackDigest is your solution. Rather than unsubscribing from valuable newsletters, use our AI-powered summarization to get the best content from all your subscriptions in one digestible format. You can use semantic search powered by vector similarity to run digests on specific topics, adjust article scoring to prioritize what matters most, and organize newsletters into collections. This way, you maintain access to all your subscriptions while actually reading the content that's most relevant to you.

StackDigest is built for people who subscribe to many newsletters; I personally tested it with 200+ subscriptions! It's designed to help readers manage large collections efficiently by using smart article scoring to surface the most engaging content, AI summaries to condense long reads, and semantic topic filtering to focus on your interests. You can create different collections for different topics and generate separate digests for each.

Newsletter digests use a balanced scoring model that evaluates articles based on two main factors: article length (50% weight) and engagement (50% weight).

Length scoring uses progressive tiers: articles under 500 words receive lower scores (0-30%), articles between 500-2,000 words are in the optimal range (30-80%), and longer articles up to 5,000+ words receive the highest scores (80-100%). This ensures you get comprehensive, well-developed content.

Engagement scoring combines comments and reactions with similar progressive scaling: 0-10 engagements (0-40%), 10-40 engagements (40-80%), and 40+ engagements (80-100%). A small recency bonus (up to 0.1 points) is added to favor newer articles and break ties.

You can customize these weights in your digest settings if you prefer to prioritize length or engagement differently.

Semantic digests use a hybrid scoring model that prioritizes relevance to your search query while still ensuring quality content. The scoring has three components:

1. Semantic relevance (70% weight): Articles are scored based on how closely they match your search query using AI embeddings. Articles below your minimum relevance threshold (default 40%) are filtered out entirely.

2. Article length (20% weight): Length is normalized against all articles in the result set, with progressive penalties for very short content: articles under 100 words get a 90% penalty, 100-200 words get an 80% penalty, 200-400 words get a 50% penalty, and 400-600 words get a 20% penalty. Articles over 600 words receive full length scores.

3. Engagement (10% weight): Comments and reactions are normalized against the highest engagement in the result set, giving a small boost to articles that sparked discussion.

This 70-20-10 split ensures that semantic digests surface articles highly relevant to your interests while filtering out shallow content that lacks substance.