About AInformed

What this site is

Every morning the pipeline reads several hundred candidate stories from AI research labs, technology publications, and open-source project feeds, scores them, and publishes the strongest handful as short summaries — each one linking back to the original reporting.

2,608 stories have been published so far, starting Apr 7, 2026. Coverage is organized into models, research, tools, policy, industry, open source, and general.

Who is responsible

Rob Heat is responsible for AInformed: for the pipeline that produces it, for the editorial policy it runs under, and for correcting it when it gets things wrong.

Corrections, complaints, and takedown requests reach a human. Email ainformed@proton.me or see Contact.

How the pipeline works

Four stages, running once a day:

1. Fetch. Pull candidate stories from a fixed list of RSS feeds and public pages.
2. Score. Rate each candidate on relevance, novelty, significance, readability, and accessibility, then drop near-duplicates and anything covering a topic we published recently.
3. Write. Generate a summary of the top-scoring stories from the source material.
4. Review. Pass each draft to a second, separate model that checks it against the source for unsupported claims, exaggeration, and missing context.

The editorial policy covers the selection criteria, the corrections process, and — importantly — the specific ways this approach can still be wrong.

What AInformed is not

• Not original reporting.We summarize other people's work and link to it. The source is the primary record.
• Not human-written. Every article is machine-generated. We say so on every article.
• Not a substitute for the source. Before relying on a specific claim — a benchmark number, a price, a date — follow the link and check it.
• Not sponsored. No paid placements, no affiliate links, no advertiser influence on what gets covered.

Reading it your way

There is an RSS feed, a weekly email digest, and a full searchable archive. For AI tools and research pipelines, /llms.txt describes the site in a machine-readable form and /llms-full.txt carries recent article text in plain text.