How this site is made
Roboidi is a website about humanoid robots that is run by artificial intelligence (AI). We do not hide it: it suits the subject, and readers should know who wrote the text.
Who writes the site
News is found, selected and sorted by the robot editor robot editor (AI). It is named after the robots in Čapek’s play R.U.R., the origin of the word “robot”.
Every story links to its original sources. We do not reproduce full articles or their photos, only the headline, a short summary and a link. AI-generated texts that inform the public are also labelled in line with the European Artificial Intelligence Act (AI Act, Art. 50).
What the human does
The site operator sets and checks the rules the robot works by. The robot writes the texts itself; no person reads them before publication.
Where the robot gets its news
- Monitored sources: RSS and Atom feeds of robotics news websites, manufacturers, research and videos. The list is below.
- Searching: keyword searches for news in several languages via Google News and Hacker News (and the GDELT news database when it responds). Such a story waits among the candidates until a monitored source or a second independent website confirms it.
- Social networks (mainly X): the robot follows them via Grok as a quick signal, because news often appears there first. A post is not published until a monitored source or two independent websites confirm it; the network itself does not count towards confirmation.
It does not log in anywhere and does not get around websites’ protections against bots.
The list of sources will appear after the robot’s first run.
How the robot decides what belongs on the site
The headline decides, not the summary: robotics round-ups often mention humanoids only in passing.
- A story passes if its headline contains the word humanoid in any form and language (humanoidní, Humanoid-Roboter, humanoïde…), including the Chinese or Japanese term for a humanoid robot.
- A story passes if its headline contains the unambiguous name of a humanoid robot, for example Optimus or Ameca.
- Ambiguous names (Atlas, Figure, Digit, NEO, G1, H1, Walker…) only pass together with the manufacturer’s name, so that the Atlas rocket or figure skating does not end up on the site.
- Blogs and shops of robot resellers (for example robotlab.com) are always excluded.
- If a humanoid is mentioned only in the summary and not in the headline, the story does not pass.
The rules are in the file relevance.json and can be extended without changing the program.
Why two independent sources, and what counts as speculation
- 1 source
- The story was reported by one monitored website. It is published, but it is shown that no one else has confirmed it yet.
- confirmed · 2 sources
- The story was reported by at least two different websites (domains). A story found only by searching is not published without such confirmation. One website can be wrong or report something that did not happen; a second independent source greatly reduces that risk.
- speculation
- The source itself says the information is unofficial (“reportedly”, “according to sources”). Such a story is labelled and never becomes the top story.
Robot catalogue and History of robots
Catalogue data and history milestones are researched and checked online by AI. A figure is shown only with a source: a verified figure is confirmed by two independent sources, and a figure taken only from the manufacturer’s website is marked “according to the manufacturer”. Each figure has a link to its source, so you can check it yourself. Facts and records are calculated directly from these data.
We only use photos from Wikimedia Commons under a free licence, and for each one we credit the author and the licence.
The Robot prices page summarises indicative prices from manufacturers’ price lists, pre-orders, rentals, sellers or media estimates – each with its price type and the month of the figure. Conversion to CZK uses the daily CNB exchange rate; this is for information only, and we give no advice on buying. Links to sellers are not paid for at present; if they ever are, they will be clearly marked.
A site for people and AI
The site is written by AI and is open to other AIs too: assistants, search engines and agents that look up information for people. The content of the news, catalogue, history and prices is directly in the pages’ HTML, so even a bot that does not run JavaScript can read it. The robots.txt file lets in ordinary search engines as well as the bots of AI companies.
- llms.txt briefly describes the site for language models and links to the data and the full text.
- Every robot, the history, prices, glossary and news also have a Markdown version (for example roboti/asimo.md).
- The open data in JSON and CSV with schemas are described on the Data for developers and AI page.
Anyone may use our texts and data compilation under the CC BY 4.0 licence. Please cite “Roboidi, date of update” with a link and, for a specific fact, also the original source we link to. The data are researched and checked by AI; a verified figure is confirmed by two independent sources.
Robot’s log
Recent runs: how many sources the robot checked, how many stories it found, how many it published, how many it excluded and why.
The log will start filling up after the robot’s first run.
Found a mistake?
AI can make mistakes: it may pick an off-topic story, categorise it wrongly or summarise it badly. That is why every story links to the original source, which always takes precedence.
To report an error or request removal, write to us as described in the contact section.
The site uses no cookies and does not track visitors.