#How it works

Follow Anywhere uses one mental model: give it something you care about, turn that into followable sources, and pay attention when something meaningful changes.

#1. Start with whatever you already have

The homepage accepts a URL, a topic, a list of URLs, pasted OPML/XML, or a dropped .opml / .xml file.

For OPML, parsing is deterministic: Follow Anywhere walks the outline tree, extracts source URLs, keeps folder context where possible, normalizes URLs, and removes duplicates before showing a review preview.

#2. Resolve what each input means

A source may already be an RSS/Atom feed, it may be a normal webpage, or it may be a supported platform URL that the Web Feed service can resolve. A topic is different: it represents an interest that may require multiple sources.

The collection mechanism is an implementation detail. Users should not have to decide which scraper, API, or feed format sits underneath a follow object.

#3. Add intelligence after collection

AI is most useful after deterministic parsing and collection: classify sources, find canonical feeds, group related sources, rank relevance, deduplicate repeated stories, and summarize meaningful changes.

The useful unit is not every fetch or every tiny DOM mutation. It is an update worth the user's attention.

#4. Keep the output open

Follow Anywhere is the product layer. RSS and JSON Feed remain useful compatibility formats for integrations and existing readers.

#Why OPML matters

OPML gives experienced RSS users a low-friction migration path: bring an existing subscription graph in first, then gradually add webpages and topics that traditional feed readers cannot represent well.

That makes Follow Anywhere broader than an AI feed reader without abandoning the open ecosystem that makes feeds valuable.