About ClusterWire

An AI research desk for crypto news

ClusterWire.news is an automated crypto research desk built to reduce noise. It continuously reads crypto news sources, groups related stories together, and publishes short, cited briefs so you can understand what’s happening without juggling twenty tabs. Everything here is generated by software — with safeguards — but it’s still AI, and mistakes are possible. Always do your own research. This project exists because crypto news is overwhelming, fragmented, and often repetitive. ClusterWire tries to surface the signal, not the hype.

What it is (and what it isn’t)

ClusterWire.news is a solo hobby project. It’s not a newsroom, not a trading signal, and not an oracle of truth.

There are no guarantees of completeness, correctness, or coverage. The system actively tries to block incomplete or weakly sourced stories, but it won’t catch everything. When something looks off, skepticism is the correct response — always.

If the site helps you stay oriented in the crypto space, that’s success.

How it works

From raw signals to cited summaries

ClusterWire follows a structured pipeline designed to keep automation honest and traceable.

  1. 01

    Signal intake

    At the moment, the system monitors selected crypto news sources via RSS feeds. Each item is stored with its source and metadata so it can always be traced back.

    More input channels (such as filings, on-chain events, or developer updates) may be added over time, but the core principle stays the same: clear provenance first.

  2. 02

    Enrichment & clustering

    Incoming items are analyzed and grouped when multiple sources describe the same event.

    These clusters let you see corroboration side-by-side instead of relying on a single headline.

  3. 03

    Draft generation

    The system assembles short articles that summarize what happened, reference the underlying sources, and highlight uncertainty when details are incomplete.

    Different events call for different tones — some are straightforward news, others need more context — but the goal is always clarity over drama.

  4. 04

    Quality checks & publishing

    Before anything is published, automated checks look for missing sources, weak coverage, and structural issues. Some items are blocked or delayed when the information isn’t solid enough.

    Despite this, errors can still slip through. Automation helps scale attention, not eliminate responsibility.

Ingestion

RSS

Reg filingsOn-chain alerts (planned)

Understanding

Embeddings

Entity graphsSimilarity

Narration

Draft builder

Citation weaverStyle coach

Trust gates

Factuality score

Source minimumsHuman overrides
What makes it useful

Evidence over velocity

Speed matters, but only when claims can be traced back to real sources.

Transparent lineage

Every summary is built from visible inputs. You can verify the original material in seconds.

Less repetition

If ten outlets report the same event, you shouldn’t have to read all ten.

Calm by design

No influencer takes, no urgency theater. When price predictions matter, we report who said what and link directly to the source. Just "here’s what happened, and here’s where it came from."

Why I built it

I built ClusterWire because following crypto felt like constant tab juggling — duplicate headlines, partial information, and noise hiding the few things that actually matter.

This is the tool I wanted to exist.

This project started as an experiment — partly technical curiosity, partly frustration — and turned into something I now use myself. I’m sharing it publicly in the hope that it helps others stay informed without burning attention.

Support & sustainability

ClusterWire.news is run independently and funded out of pocket.

If you find it useful and want to help keep it running, crypto donations are appreciated. They go directly toward hosting, infrastructure, and keeping the system alive.

No paywalls, no partnerships, no promises — just a tool, shared openly.

Privacy note: we log tip UI events (page + action, and article slug when applicable) to improve the feature. We don’t store IP address, user-agent, or wallet addresses in analytics. Tips are on-chain, so the sending address is public in the transaction.

Questions, feedback, or bug reports are always welcome — especially when something looks off.