How Crada triages your content before you read it

The whole point of Crada is not saving content — it's knowing what's worth reading before you invest your time. Every piece of content that enters Crada goes through a triage engine that answers one question: does this deserve your attention?

Here's how it works and why it works the way it does.

Three layers of triage

When content arrives in Crada — from Gmail, the Chrome extension, the bookmarklet, or a manual save — three things happen automatically.

1. Novelty scoring against your knowledge base

This is the core differentiator. Every new article is compared against everything you've already ingested. Not just deduplication — Crada asks: does this bring arguments, data, or perspectives you haven't encountered yet?

The result is a novelty score. A high score means genuinely new material. A low score means you've seen these arguments before, repackaged by a different author.

This matters because the most dangerous time waster isn't bad content — it's good content you've already absorbed. Without novelty scoring, you end up re-reading the same insights framed slightly differently, thinking you're learning when you're just pattern-matching.

2. Intent detection

Not all content is created equal in purpose. Crada evaluates the narrative intent behind every piece:

• Substance — genuine analysis with original arguments or data
• Sales pitch — content designed to sell a product or service, disguised as insight
• Hype — riding a trending topic with no real depth underneath
• Recycled take — the same argument you've read before, rewritten by someone else

This is the filter your instinct provides when you scan headlines — but automated and consistent. You no longer need to read three paragraphs to realize an article is a thinly veiled product pitch.

3. Signal, Momentum, Background

After analysis, every piece of content gets a verdict:

• Signal — genuinely novel developments that should change your strategy or understanding. These earned your time.
• Momentum — themes gaining traction across your field. Worth tracking, not urgent.
• Background — useful context, but nothing you haven't seen before. Skip with confidence.

Open Crada and go straight to Signals. That's your reading list, pre-filtered by an AI that knows what you've already read.

Your categories, refined by you

Crada doesn't use generic tags. The AI suggests thematic categories based on the content — and when you edit them, every future article benefits from that refinement.

Say one of your themes is "AI impact on data infrastructure." When new content arrives, the AI checks it against your existing themes and categories. It either places it in an existing bucket or suggests a new one. You can accept, edit, or override.

Over time, the system classifies content exactly the way you think. Not generic labels imposed by an algorithm — your taxonomy, built through use.

Full-text search underneath

Triage is the primary workflow, but search is the backbone. Every piece of text gets indexed with weighted ranking:

You can search naturally: AI infrastructure finds items containing both terms. "autonomous agents" uses exact phrase matching. react -native excludes React Native results.

What's coming: chat with your knowledge base

We're building a conversational interface where you can ask questions about your content library. Think of it as a ChatGPT grounded in the articles and insights you've actually vetted and saved — not the entire internet.

This turns Crada from a triage tool into a research companion. Ask a question, get answers sourced from your curated knowledge base.

Triage is not an afterthought

Most tools save first and analyze never. Crada is built the other way around. Analysis happens the moment content arrives. Every feature we add has to make triage better — or at least not get in the way.

Save content. Get a verdict before you read. Go straight to what matters. That's the promise, and the triage engine is how we deliver on it.