Our reader
WelilaBot reads the public sources an assistant would, identifies itself, signs what it sends, and stops the moment a site asks it to.
To keep a business's facts current, Welila reads the same public sources an assistant would: the website, the search listings and the directories. The reader that does this is WelilaBot. It reads pages only. It never submits a form, signs in, or fetches anything behind a login. This is the page a bot directory asks for, and the page to find if you see WelilaBot in your logs.
- Name
- WelilaBot.
- Operator
- Welila, a trading name of Hoosei Limited, United Kingdom.
- What it does
- Reads the public website and public listings of a claimed business to keep its confirmed facts current: hours, services, prices, credentials and location.
- How often
- Once a month for a claimed business, and again when the business asks. Never more than one request a second to a host.
- How to recognise it
- A user agent that names WelilaBot and links to this page. Every request is signed, and the public keys are at /.well-known/http-message-signatures-directory, named in the Signature-Agent header.
- How to block it
- A robots.txt Disallow for WelilaBot, honoured on the next read. It honours robots.txt and crawl-delay already. You can also write to hello@welila.com.
- What it stores
- The facts it reads, the source and the date it read them. Not the page itself, and never personal information.
- Contact
- hello@welila.com.
The assistants Welila reads for
Welila serves the assistants people actually ask: ChatGPT, Claude, Perplexity, Google and Bing. The robots policy allows the readers those operators publish and refuses the two bulk scrapers that train without reading for a person.
If you run a site
You do not have to do anything. WelilaBot takes what is public and takes it gently. If you want it to slow down or stop, publish the rule and it is honoured, and you can write to hello@welila.com.