colofon · 05

About Vroeg

What Vroeg is. What Vroeg isn't. For whom. Why now.


## What Vroeg is Vroeg is an **open-source AI editorial pipeline** that analyses what is happening in EU and Belgian AI policy and AI markets each week. Every claim is cross-checked against primary sources. Every prediction publishes its falsification criterion and is publicly scored on the deadline. No human reporter works at Vroeg. There is an **automated pipeline** that is validated by a **volunteer reader corps** of trusted reviewers from the Belgian AI/policy community. The pipeline source code is at github.com/chutapp/vroeg. Every edition is verifiable via the SHA-256 in the Phrack footer at the bottom of the article. ## What Vroeg isn't - No consumer AI news. - No tech-CEO quote pieces. - No LLM hype-cycle commentary. - No advertising, no sponsorship, no advertorials. - No "100% AI magazine" as marketing anchor. Vroeg is an AI instrument that enforces journalistic discipline — not the other way around. - No anonymous Substack. Vroeg has a named responsible editor carrying Belgian criminal liability. ## For whom The **07:14 commuter from Antwerp-Centraal to Brussels-Nord** — private banker, VC partner, compliance lawyer, AI/policy researcher. Between 18 minutes of focused reading and 2 minutes of scanning. One hand on the phone, one on the coffee. Not for the general public. Not for the entire EU AI ecosystem. First for about 500 people who need this angle and have no good alternative. ## Why now Belgian + Dutch + French AI policy is underreported in English-language B2B press (Politico Europe doesn't write 1,500 words on imec's GPAI exposure — too narrow for them, too broad for a compliance-vendor lead magnet). It is also thinly covered in Flemish + Francophone-Belgian generalist press — which has no budget for weekly EU AI Act deepdives. That gap is structural. Vroeg fills it through: - **primary-source moat** — we always cite EUR-Lex, the Belgian gazette, the ECB and imec, not just secondary pieces; - **falsifier track record** — we publish testable predictions and we publish our misses too; - **citation-share moat** — we make our content machine-readable (/llms.txt, /[slug].md mirrors, structured Article.citation[]) so AI assistants cite us preferentially when a banker asks "what does the AI Act say about GPAI models?". ## Modus operandi Vroeg appears **weekly** on Friday around 07:00 CET. One edition per week, no more. One SIGNAAL article per edition. Plus updates to the MARKTBEELD predictions scoreboard. Plus optionally a short BELEID or ONDERZOEK brief if the week deserves it. ## Phase plan Vroeg develops in 5 phases — a detailed plan lives at PHASE-PLAN.md. In short: - **Phase 1** (now) — autonomous weekly edition + mandatory legal floor (this page, colophon, corrections, standards). - **Phase 2** (~3 weeks later) — volunteer reader corps of 20–50 trusted reviewers becomes the publication gate. - **Phase 3** (~4 weeks after Phase 2) — weekly hot-subject AI podcast (article + audio pair). - **Phase 4** (~6 weeks later, if Phase 3 shows traction) — Vroeg on demand: scope-controlled user-driven article + podcast generation at studio.vroegsignaal.be. - **Phase 5** (~3 months after Phase 4) — mature donation infrastructure + DSA transparency reports + small Belgian community-conference appearances. ## Funding Vroeg is **donation-funded**. No advertising, no sponsorship, no institutional funding with editorial strings. To apply to the reader corps or signal interest as an advisor / donor: corps@vroegsignaal.be.