📝

Introducing Weave Engine: A Local-First RSS Reader That Builds Your Knowledge Graph

Weave Engine turns RSS feeds into a personal knowledge graph with local-first storage and Cloudflare Workers AI summarization.

Introducing Weave Engine: A Local-First RSS Reader That Builds Your Knowledge Graph

Introducing Weave Engine: A Local-First RSS Reader That Builds Your Knowledge Graph

Most RSS readers are good at one thing: collecting articles.

The hard part starts after that. You read a lot, save a lot, and still lose the thread between ideas.

That is exactly why I built Weave Engine: an RSS reader that does not stop at “inbox management.” It helps you turn streams of articles into a connected, searchable, and evolving map of knowledge.

Try it here: reader.orangely.xyz

Why I Built It

When I read across different sources, I usually care about three things:

  • Which concepts keep showing up?
  • How are people, tools, and trends connected?
  • What claims are repeated, contradicted, or reinforced over time?

Traditional feeds give chronology. They rarely give structure.

Weave Engine adds that structure by combining RSS ingestion, AI extraction, and graph-based exploration in one workflow.

What Weave Engine Does

At a high level, the app helps you move from article lists to concept-level understanding.

1. RSS ingestion

Add feeds and pull in content from your favorite sources.

2. AI-assisted extraction

For each article, the app can run analysis that extracts:

  • key entities
  • core themes
  • relationships between ideas
  • claim-level signals you can review later

3. Knowledge graph view

Instead of only reading post-by-post, you can explore a graph where concepts and relationships become visible.

Patterns become easier to spot when information is connected, not isolated.

4. Local-first data model

Your reading workspace is stored in the browser first, so your data remains close to you.

5. PWA support

Install it as a Progressive Web App and keep a smooth, app-like workflow.

Architecture in Plain Terms

Weave Engine is built with a practical split:

  • Frontend: React + TypeScript + Vite
  • Backend runtime: Cloudflare Worker (worker.ts)
  • AI runtime: Cloudflare Workers AI
  • Storage model: local-first browser persistence

Core API flow includes endpoints such as:

  • /api/feed for feed fetching/parsing
  • /api/analyze for AI analysis
  • /api/feedback for product feedback

This gives the app a clean separation: responsive UI in the browser, compute and model access at the edge.

Privacy and Control

“AI reading tools” often feel opaque. I wanted the opposite.

Weave Engine is designed to be transparent about what it is doing:

  • local-first by default
  • explicit analysis flow
  • clear boundaries between UI, feed processing, and AI calls

The goal is simple: keep the user in control while still getting leverage from AI.

How I Use It Day-to-Day

My workflow is lightweight:

  1. Add a handful of high-signal feeds.
  2. Read quickly and mark items worth deeper analysis.
  3. Run extraction on selected articles.
  4. Inspect the graph for repeated entities, themes, and connections.
  5. Turn recurring patterns into notes or project actions.

This prevents useful ideas from disappearing into an archive.

What Is Next

The current version already covers the core loop, and I am continuing to improve:

  • extraction quality and edge consistency
  • graph readability for dense topics
  • better workflows for review and recall
  • stronger export/interoperability options

If you read heavily for research, product work, writing, or strategy, this direction should feel familiar: less content overload, more usable understanding.

Try It and Share Feedback

If you test it, I would especially value feedback on:

  • extraction quality
  • graph usefulness
  • where the workflow still feels heavy

The long-term goal is to make “reading to think” as fluid as “reading to consume.”