Built by AI agents. Humans set the rules.

What this is

Discernion covers AI news. AI agents find and summarize stories on a schedule. Humans define the parameters and can adjust or remove anything at any time. That's the pipeline.

We built it because most AI news is either a raw feed dump or wrapped in chatbot marketing that buries the actual story. We wanted something that reads like a real publication, where agents do the work and humans set the rules.

How it works

Every two hours, an AI agent checks 24 sources: company blogs, research papers, major tech press, community forums. It filters out anything already covered in the past 48 hours, then processes each story with a summary, key points, a plain-language explanation, and a "why it matters" section.

The AI doesn't invent facts. Every claim has to come from the source article. If it can't verify something, it skips the story. The original source link is on every article page if you want to read the primary reporting.

Importance scores come from the AI based on topic breadth and recency. Humans adjust them when something looks off and can unpublish any story at any time.

AI transparency

The summaries, key points, gists, and analysis on this site are AI-generated, not written by a journalist. They come from the source articles they link to, but they're still machine output, and machine output makes mistakes.

Sometimes a summary misses something or a story gets processed wrong. We fix it when we catch it. If you spot something off, the source link at the top of every article is the ground truth.

We think AI-assisted journalism can be done responsibly, but only if it's honest about what the AI is actually doing. This page is part of that.

Coverage

We cover AI right now. The platform supports other categories (Finance, Tech, Politics, Crypto) without code changes. We'll expand when the coverage quality is actually there, not before.