RelevaBot reads the pages a store asks its sales agent to know.
If you found RelevaBot in your logs, it was sent by an online store that uses Releva. This page explains what it reads, how often, and how to allow or stop it.
How to recognise it
RelevaBot identifies itself in the User-Agent header of every request. When it opens a page in a browser it sends:
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; RelevaBot/1.0; +https://releva.ai/bot) Chrome/141.0.0.0 Safari/537.36
The Chrome number follows the browser version RelevaBot runs, so it changes over time. When it downloads a document, such as a PDF, it sends:
Mozilla/5.0 (compatible; RelevaBot/1.0; +https://releva.ai/bot)
Both contain the same token, RelevaBot/1.0 followed by this page's address. Match on that token, not on the whole string.
What it reads, and why
Releva gives an online store an AI sales agent that answers shoppers' questions. To answer about returns, delivery, warranty, sizes or payment, the agent needs the store's own pages on those subjects. The store adds those pages to the agent's knowledge, and RelevaBot fetches them. It also reads a store's home page when the store asks Releva to describe its brand and tone of voice.
RelevaBot reads only the addresses a store has added. It does not follow links and it does not crawl a site. The text it reads is used only by that store's own Releva agent.
How it visits
RelevaBot opens a page in a headless Chrome browser, so pages built with JavaScript show their real content. It waits a few seconds for the page to finish loading, and it can close a cookie banner or open collapsed sections, such as FAQ answers, when the store has set that up for its own site. If a page answers with an error, it may try again a few times in a row.
It visits when a store adds a page or asks for it to be read again, and then about once a day, to pick up changes. Each visit loads one page and the files that page needs, like a normal browser.
RelevaBot does not sign in, submit forms, add to carts or buy anything, and it does not collect information about your visitors.
Allowing RelevaBot through Cloudflare
If you are a Releva customer and your site is behind Cloudflare, Cloudflare's bot protection may block RelevaBot, and your agent then cannot learn from your pages. To let it through, open your domain in the Cloudflare dashboard, go to Security, then WAF, then Custom rules, and create a rule with the expression (http.user_agent contains "RelevaBot") and the action Skip. Under the action, choose to skip Super Bot Fight Mode and, if you use them, the managed rules and rate limiting rules. Place the rule first.
On Cloudflare's Free plan, Bot Fight Mode cannot be skipped by a rule. If it is on, switch it off (Security, then Bots) or move to a paid plan to use the rule above. Other firewalls and security plugins have a similar allow list: add the RelevaBot token there.
A User-Agent header can be copied by anyone, so a rule on the token alone also lets through anyone pretending to be RelevaBot. Keep the rule to the pages your agent needs if your firewall allows it.
Stopping RelevaBot
If you are the store that added the page, remove it from your agent's knowledge in the Releva dashboard and RelevaBot stops visiting it. If you run a site that RelevaBot visits and you did not expect it, write to support@releva.ai with your domain, and we will stop it.