Newsroom & method

The i6eal newsroom publishes original AI journalism for Germany. Here's who's behind it, how we work, and what you can rely on.

32sources worldwide — labs, research, trade press
hourlyreview of the fresh news picture
3automated checks before any release
100%of articles editorially reviewed and released

How a story gets made

Hundreds of candidates a day become a few, verified articles. Every story passes through this chain — and nothing goes live without a human sign-off.

Two source streams

No story appears out of nowhere. The newsroom is fed by our own measurement tools — AI-crawler monitor, latency index, price trackers — and 32 curated external sources: from the AI labs' blogs to arXiv and the trade press.

OpenAI · Google DeepMind · Anthropic · arXiv · The Decoder · heise · Reuters …

Hourly review

Every hour we look at the fresh news picture. A memory keeps track of everything already seen — repeats and reheated stories don't even get through the door.

On a typical day: ~150 candidates — most of them duplicates.

Editorial ranking

A strict editorial rule set selects: only concrete developments, only with at least one reputable source. PR fluff, blog advertising and no-name portals are thrown out.

~150 candidates become 3–6 articles. On quiet days: zero.

↳ PR & noise → discarded

Draft

The article is written bilingually (DE + EN), with linked sources and its own cover image. Quotes are never invented — if it's not in the source, it's not in the article.

Fact-check

Every factual claim is checked against the source material: numbers, names, quotes — even whether the headline is phrased harder than the source. The result is a fact-fidelity score from 0 to 100.

Real example: “Samsung manufactures AI chips for Anthropic” → stopped. The source only mentioned talks.

↳ unsupported → stopped

Revision

Flagged passages are tightened or removed — and the article is then re-checked in full. If it doesn't get better, it stays on the bench.

The newsroom decides

No article goes live automatically. We review every piece ourselves, sharpen it where needed — and decide what carries the i6eal name. Only what convinces us gets published.

Publication

Approved articles appear spread across the day — with source citations, a named author and a corrections notice on every article.

What you'll find here

The newsroom stands on two legs. First, our own data analysis — first-hand AI journalism grounded in measurement tools i6eal runs itself: the daily reading of AI-crawler blocking across Germany's 1,000 largest websites, or AI latency measured from Frankfurt. These pieces link to the matching live tracker so you can check the raw data yourself.

Second, the daily AI news beat: we track the sources that matter — AI labs, research, and the trade press — and put the most relevant developments into context for a German audience. Those pieces name and link their sources openly; we never pass off someone else's reporting as our own.

How we work

Every data-driven analysis includes a method section: we name the data source, the collection window, and the limits of the measurement. Where we estimate, we label it an estimate; where we cite, we link the primary source.

Our news pieces are drafted with AI support from openly named sources and are automatically checked against that source material. A named human is accountable: Ideal Syka. We'd rather publish a few carefully checked pieces a day than automated volume.

Corrections

Mistakes happen. When we make one, we correct it visibly and note the change with its date on the article. If you spot something, tell us — we check every report.

Independence

i6eal is an AI studio, not a neutral publisher — we say so openly. Our analyses are nonetheless as honest as our data allows: we publish a number even when it's inconvenient. Where an article touches one of our own products, we disclose it.