NOLOBot, the visitor sent by a person.
NOLOBot is how NOLO identifies itself when it opens a web page for one of its users. It is not a crawler: it visits only when someone asks NOLO something that needs your page, it follows your robots.txt, and you can turn it away in two lines.
When it visits
Reading pages. When a user pastes a link into a chat, or NOLO needs what a page says to answer (a search result, the results page of your site's own search, or a link on a page it already read), NOLO's server downloads that page and reads its text, up to ten pages for one answer. It identifies itself as Mozilla/5.0 (compatible; NOLOBot/1.0). On the Pro plan, a page that only shows its content to a real browser, because it is built with JavaScript or because your site turns plain readers away, is opened with the browser described below, without clicking or typing anything, and closed right after.
Browsing for a task. On the Pro plan, when a user asks NOLO to do something on a website, such as checking a price behind your search form or how your site looks on a computer and on a phone, NOLO uses a real web browser on its own server. It is an ordinary Chrome whose user agent ends in NOLOBot/1.0 (+https://landing.nolo-app.com/bot/). It clicks, scrolls and types like a person would, one task at a time, and NOLO's whole service runs at most two of these browsers for tasks at once, plus two more that only open pages to read them. Each task starts with a fresh browser, with no cookies or stored data, and everything is deleted when it ends: no cookie your site sets survives into another visit. When a user asks how your site looks on both a computer and a phone, the phone view loads each page the computer view opens, so you may see two visits to the same page.
Both come from NOLO's own server in France. Neither indexes or crawls your site, or keeps a copy of your pages: they open only what one person's question needs, and what they read goes into one answer for that person.
It follows robots.txt
Before it opens a page, NOLO reads your robots.txt and does what it says (RFC 9309). It looks for a group named NOLOBot and, if there is none, follows the * group. It also treats a group for the readers of other AI assistants (Claude-User, ChatGPT-User, Perplexity-User) as meant for it: such a group can only close pages to NOLO, never open what * closes. A robots.txt that does not answer (an error, a 429, a timeout) closes the whole site until it answers again.
To keep NOLOBot off your whole site:
User-agent: NOLOBot
Disallow: /
Or only part of it, for example Disallow: /account/. NOLO reads a site's robots.txt again every few hours.
What it never does
- It never logs in, creates accounts, or types passwords, card numbers or one-time codes. Those fields are refused in its code, not just in its instructions.
- It never pays or starts a checkout.
- It never solves CAPTCHAs or "verify you are human" checks: it stops and tells the user.
- It never uploads or downloads files, and never posts, comments or reviews.
- It never sends a form on its own. If the user asked for a contact form or a booking request, NOLO fills it in and shows it to them; it is sent only if they tap Send.
- It never reaches private addresses: every connection is checked against public internet addresses, on ports 80 and 443 only.
Questions or problems
If NOLOBot does something on your site that it should not, or you would like us to know about your site's limits, write to nolo.ai.app@gmail.com. It reaches a real person.
More about what leaves NOLO and where: Transparency and the privacy policy.