How content is produced
For each article, a language model writes the title, body and metadata from a topic and a set of common questions. It works from prompts our team writes and reviews. Before publishing, the draft goes through automated checks: structure, length, required fields. Someone on the team can also review and adjust an article, but not every article is read by a person before it goes live.
Sources
Since September 2026, we check every specific claim in every article against outside sources, in whatever language has the best one. That covers figures, units, ranges, records, named species and places, taxonomy, and conservation status.
We compare claims against sources like accredited zoos, university databases such as Animal Diversity Web, and the IUCN Red List. Depending on the topic, we also draw on peer-reviewed research or established references like National Geographic and Britannica.
Each article ends with a Sources section listing the pages that back its facts. When we can’t find a source for a figure, we correct it, make it more general, or remove it. If a question doesn’t have a published answer yet, we keep the commonly quoted figure as an approximation and mark it unverified.
Some questions, like which of two animals would win a fight, have no recorded answer. For those, an article can end with an “Our estimate” section. It is our own inference, built only from the figures the article cites, with the math shown, and it says so at the top. It is never presented as a documented result.
The institutions we cite don’t review or endorse our articles.
Corrections
Found an error? Email [email protected] with the article title or URL and what’s wrong, and we’ll review and fix it.
Images
Images on this site are AI-generated illustrations, not photographs.