Call · 30 min
MarketingMattia Esposito10 November 202613 min read

How to appear in Mistral Le Chat, now Vibe. Mistral searches an index of its own, and the crawler to let in is called MistralAI-Index.

Le Chat changed its name in May and, to answer, Mistral looks things up in an index it fills with a crawler of its own. Here is which crawler to let in and a checklist to verify it, in Word and PDF.

In brief

To appear in Vibe, the Mistral assistant called Le Chat until 28 May 2026, your site has to let in MistralAI-Index. Mistral uses this crawler to index the web for Mistral search, the search that helps Vibe answer.

Mistral documents three crawlers with three jobs. MistralAI-User opens a page when a user asks for it; MistralAI-Training is for training and can be blocked without dropping out of the answers. For news, Vibe uses AFP and Associated Press.

Once inside the index, a readable page is what counts. Text in the HTML, no login, a precise answer and facts you can check: Vibe shows every source in its Sources panel.

The checklist with the robots.txt lines and the test can be downloaded below, in Word and PDF, with no email required. Updated on 10 November 2026.

Written by: Mattia Esposito, AI and marketing consultant, founder of Itria AI. From Puglia, based in Turin since 2023, he works remotely in Italy and abroad. LinkedIn profile.

The Mistral Vibe checklist, to download

One sheet with Mistral’s three crawlers, the robots.txt lines to copy and nine checks with room for the result, in Word to fill in and in PDF to print. You need access to robots.txt, to the server or CDN logs and a free Vibe account. The first five rows take half an hour.

The Mistral Vibe checklist, formerly Le Chat: Mistral’s three crawlers, the robots.txt lines and the test in nine checks
FileContentsLink
Wordto fill in, in Italian

The three crawlers with your choice, the robots.txt lines, the nine checks with their result.

come-comparire-su-mistral-le-chat-scheda.docx

PDFto print, in Italian

The same checklist, to tick off with a pen.

come-comparire-su-mistral-le-chat-scheda.pdf

N.What to checkWhere to look
1

Your robots.txt does not block MistralAI-Index or MistralAI-User, not even through a general block.

yoursite.com/robots.txt

2

Your CDN or firewall does not stop Mistral’s crawlers before robots.txt is read.

CDN or hosting dashboard

3

Visits from MistralAI-Index appear in the logs, with the date of the latest.

Server logs

4

The IPs of those visits are on the lists Mistral publishes.

mistral.ai, JSON files of the IPs

5

Text, prices and contact details sit in the HTML, with no login and no script.

Page source (Ctrl+U)

6-8

Three buying questions in Vibe with Web search on: does the Sources panel show your site?

chat.mistral.ai

9

chat.mistral.ai appears among the referrers of your visits.

Site analytics

On 7 August a friend sent me what ChatGPT was saying about me: it couldn’t find a Mattia Esposito linked to itria.io as its founder. Since that day I have checked, assistant by assistant, which door each one comes in through. For ChatGPT, Copilot and Google the route had been documented for months. For Mistral I found it on a technical page in English that I had not seen quoted in any Italian guide.

On 7 September, after reading the websites of 142 Italian companies that also sell abroad, I wrote: “An assistant doesn’t browse a site: it reads it, and reports what it finds written there.” In 61 of the 142, or 43%, a foreign buyer had no real way to get in touch. Of 143 sites, 28 had the file that explains the company to assistants, and in 24 a plugin had written it. The full post.

Vibe is no exception. The rest of this page sets out what Mistral has published, with the date I read it, and what no source says yet.

Mistral Le Chat is now Vibe: what changes if you want to appear

On 28 May 2026 Mistral AI announced that Le Chat was becoming Vibe, a single agent for work and for code. Mistral’s blog speaks of “one agent and one licence across work and code”, with conversations, settings and plans carried over. The help article on the change of name is dated 12 August 2026, which is when the help page was updated, not the day of the change.

For anyone who wants to appear, little changes. The address is still chat.mistral.ai, accounts and conversations moved across with everything else, and the turn-based conversation of Le Chat lives on as Vibe Chat. Alongside it sit Vibe Work, for long tasks with tools, and Vibe Code, for programming. The web search documented today lives in Work.

FactWhat Mistral saysSource
The name

Le Chat became Vibe on 28 May 2026.

mistral.ai/news/vibe-agent

The modes

Vibe Work, Vibe Code and Vibe Chat, Le Chat’s conversation, still available.

help.mistral.ai, article 682992

The address

Unchanged: chat.mistral.ai, with the same login.

help.mistral.ai, article 682992

The index

MistralAI-Index indexes for Mistral search, not for training.

docs.mistral.ai/robots

Training

MistralAI-Training can be disallowed in robots.txt.

docs.mistral.ai/robots

News

For current-affairs questions Vibe Work uses AFP and Associated Press.

docs.mistral.ai/vibe/work

Brave

Mistral is among the declared users of the Brave Search API.

brave.com/about

The data

Hosted in the European Union by default.

help.mistral.ai, article 347629

The old name is still the one people search for. On Google, in November 2026, the Italian suggestions still say “le chat mistral”, while in English “mistral le chat vibe” has already appeared. This page uses both names for the same reason.

What is Mistral Le Chat, and why it matters for a small business

Mistral AI is the French company that develops the Mistral models and the Vibe assistant, and it matters to a small or mid-sized business because it is a European alternative to ChatGPT. A buyer in Europe who prefers a tool with data hosted in the Union can ask Vibe the same questions they would ask ChatGPT, starting with who makes something and who supplies it.

No primary source publishes market share or usage figures for Vibe in Italy, and this page gives none. The reason to care is a different one: Mistral’s door opens with two lines of robots.txt, and a site that keeps it shut by mistake stays out of one more assistant without knowing. Mistral publishes its Help Center in Italian too; the technical documentation is in English and French.

Mistral’s three crawlers and what to let through

Mistral documents three crawlers, and for getting cited the ones that count are MistralAI-Index and MistralAI-User. On the crawlers page, MistralAI-Index is described like this: “It indexes content for Mistral search, which helps answer user questions in Vibe.” Content gathered by this crawler, Mistral adds, is not used to train models of any kind.

Mistral’s three crawlers: MistralAI-Index takes the page into Mistral search and from there into Vibe’s answers, MistralAI-User opens it when a user asks, MistralAI-Training takes it into training The pagetext in the HTML MistralAI-Indexautomatic crawling MistralAI-Userrequested by a user MistralAI-Trainingmodel datasets Mistral searchMistral’s index Traininga separate choice Vibeanswer and Sources
The three routes a page can take inside Mistral, according to docs.mistral.ai/robots. The first two reach Vibe’s answer; the third goes to training and is decided separately.
CrawlerWhat it does, according to MistralWhat to do
MistralAI-Index

Crawls the web automatically, only to index content for Mistral search.

Let it in

MistralAI-User

Opens a page when a user asks Vibe a question, and cites it with a link.

Let it in

MistralAI-Training

Collects pages for the datasets used to train Mistral’s models.

A separate choice

On MistralAI-User the page chooses its verb with care: the token “governs” which sites user requests can be made to. Mistral names it as the token to use in robots.txt for those visits, and says it does not crawl the web automatically and does not gather content for training. For the first two crawlers it also publishes the lists of IP addresses, handy for spotting a fake Mistral in your logs.

In robots.txt: inside Mistral search, out of training

User-agent: MistralAI-Index
Allow: /

User-agent: MistralAI-User
Allow: /

User-agent: MistralAI-Training
Disallow: /

A robots.txt with no rules for these names already lets them in: the lines matter when there is a general block further up, perhaps copied from a template that shut out every AI crawler. The third line is your choice, and it is made separately. The names of all the other assistants, with the same rules, are in the guide to robots.txt for AI crawlers.

Some crawler directories list only MistralAI-User. A site owner who wrote their robots.txt from one of those may have opened the door to user visits and left the index outside without noticing. The check takes a minute: open yoursite.com/robots.txt and search for “Mistral”. The glossary explains what an AI crawler is.

What model and sources does Mistral Le Chat use: Mistral search, Brave, AFP

Vibe answers with what the model knows and, when the Web search tool is on, with what it finds on the web in real time. The Vibe Work documentation says every answer drawn from the web carries inline links to its sources and a Sources button that lists them all. For news, Mistral writes: “Work draws from professional news partners Agence France-Presse (AFP) and Associated Press (AP)”.

Web search is switched on from the Tools menu. With Open URL a user can paste an address and have Vibe read that page: it reads that page alone, without following links, and it cannot open pages behind a login or a paywall. A price list visible only to registered customers does not exist for Vibe.

Then there is Brave. On Brave Software’s about page the Brave Search API is described as used by “Cohere, Mistral, Perplexity, and You.com”, and in March 2025 TechCrunch recalled that in February Brave and Mistral had announced Brave’s web results inside Le Chat. I could not find the original announcement on either Brave’s site or Mistral’s.

The accurate picture today: Mistral is among the declared customers of the Brave API and since 2026 has also run an index crawler of its own. Whether Vibe still uses Brave, and for which questions, no current source says. Getting found by Brave costs little and helps with other assistants anyway: how to do it, in the guide to appearing in Brave Search.

Two cases measured on itria.io

From one page to five. On the morning of 25 September 2026, for Claude’s web search, itria.io had a single page; by the evening, after I asked Brave to recrawl seven, it had five. The robots.txt was already open: the index was missing. Before and after, with the limits of the data, in Brave and Claude.

The email address that wasn’t there. Cloudflare’s obfuscation removed the address from the HTML and put it back with a script: to an assistant, the contact page had no contact details. The same defect turned up on 14 of 121 food producers’ sites, or 12%, in the piece on the email AI assistants cannot see.

What they are not. They are not measurements on Vibe: in the documents I read, Mistral offers site owners no panel with citations. They are the two defects that shut the door on any assistant, Mistral included.

What you don’t need to appear in Mistral

No Mistral document read for this page says how Vibe chooses its sources, and none names Schema.org, FAQs or Wikipedia as criteria. Anyone presenting them as Mistral’s recipe is inferring them from other assistants. Structured data stays useful for other reasons, covered in the guide to structured data for AI, but for Mistral there is no written rule.

An agent created in Vibe under your company’s name is a tool for the person who builds it, inside their own account. It does not put your site into the answers Vibe gives other users. The same goes for the llms.txt file: in the documents I read, Mistral does not say it uses it. You can write one, as the guide to llms.txt explains, without expecting anything from Mistral.

Mistral Le Chat vs ChatGPT for anyone who wants to be cited

For a site owner, Vibe and ChatGPT work the same way: one crawler for the index, one for visits requested by users, one for training. What changes are the names to write in robots.txt and the tools for measuring. The table brings together Mistral’s crawlers page and the guide to appearing in ChatGPT, which carries OpenAI’s sources.

WhatMistral VibeChatGPT
Index crawler

MistralAI-Index, for Mistral search

OAI-SearchBot

Visits requested by the user

MistralAI-User

ChatGPT-User

Training

MistralAI-Training

GPTBot

Web search

Web search tool, switched on in Work

ChatGPT search

Sources shown

Inline links and Sources panel

Links in the answer

Measurement for site owners

Manual test: no panel in the documents I read

Visits tagged utm_source=chatgpt.com

The practical difference lies in measurement. For Copilot there is Bing’s AI Performance report, for Google there is Search Console; for Vibe, in the documents I read, there is no equivalent panel. The test is done by hand, as in the checklist, and repeated every month with the same questions. The tools for the other assistants are in the guide on how to measure GEO.

Is Mistral Le Chat private? Where the data is stored

Mistral says the data of people using its services is hosted in the European Union by default, and in the United States only through the US API endpoint. The Help Center puts it this way: “By default, your data is hosted in the European Union”. Depending on the feature, it adds, data may pass briefly outside the EU to the sub-processors listed in its Trust Center.

This is the case made for choosing Vibe over an American assistant, and it concerns the data of its users. For a site that wants to appear, nothing changes: the three crawlers and the robots.txt lines are the same in Italy and across the English-speaking world. What does change is the legal weight of the third line.

WhereWhat applies, read on 10 November 2026Source
Italy

The same three crawlers. Mistral’s Help Center is also in Italian. For training, the European rule applies.

docs.mistral.ai/robots

European Union

Data hosted in the EU by default. Disallowing MistralAI-Training is a machine-readable reservation against text and data mining.

Help Center 347629; Dir. 2019/790, Art. 4(3)

United Kingdom

The same three crawlers; the text and data mining exception covers only non-commercial research.

CDPA 1988, s. 29A

United States

The same three crawlers; data held in the United States only with the US API endpoint.

Help Center 347629

Directive (EU) 2019/790, in Article 4(3), allows text and data mining unless the rightholder has reserved it in a machine-readable way. The full picture, including the AI Act, is in the robots.txt guide. A specific case needs a lawyer: this page does not replace one.

How to check whether Vibe cites you

The most direct test is to ask Vibe your buyers’ questions, with Web search on, and look at the Sources panel. Three buying questions are enough to start: who makes what you sell in your area, who supplies it abroad, who installs it. Note the date, the language and whether your site appears among the sources.

Then the logs. Visits from MistralAI-Index tell you whether the crawler comes by and when; the IP lists Mistral publishes tell you whether it really is Mistral. In your site analytics, look among the referrers for chat.mistral.ai, Vibe’s address. If the crawler never shows up, check the CDN: some stop AI crawlers before robots.txt is read.

Finally the page. Open URL cannot open content behind a login, and Mistral does not say whether its crawler runs JavaScript: the safe choice is to keep text, prices and contact details in the HTML, as the guide on JavaScript and AI crawlers explains. The overview of every assistant, with the route from page to citation, is in the guide to generative engine optimization and in the index of marketing for small businesses.

How this is built for a real company, with the right questions to hold on each assistant, is described on the page about presence in search engines and AI assistants.

Questions and answers

How do I appear in Mistral Le Chat, now Vibe?

By letting in MistralAI-Index, the crawler Mistral uses to index the web for Mistral search, the index that helps Vibe answer, and MistralAI-User, which opens pages when a user asks. Then the page matters: text in the HTML, a precise answer, facts with their source.

Mistral’s documents offer no form to register a site with Vibe.

Are Le Chat and Vibe the same thing?

Yes. On 28 May 2026 Mistral AI announced that Le Chat was becoming Vibe, with the same account, the same conversations and the same address, chat.mistral.ai. The turn-based conversation of Le Chat is still available as Vibe Chat, alongside the Work and Code modes.

Does Mistral use Brave Search?

Mistral is among the declared users of the Brave Search API, and in February 2025 the press reported the deal for Le Chat’s web results. Since 2026 Mistral has also documented an index crawler of its own, MistralAI-Index. Whether Brave is still used, and for what, no current source says.

If I block MistralAI-Training, do I disappear from Vibe?

No. According to Mistral’s documentation, MistralAI-Training collects pages only to train models and serves neither the index nor live answers. You can disallow it in robots.txt and still let in MistralAI-Index and MistralAI-User.

How can I tell whether Vibe cites my site?

Ask Vibe three buying questions with the Web search tool switched on: the Sources panel under the answer lists every page it used. In your server logs, look for visits from MistralAI-Index; in your site analytics, look for referrals from chat.mistral.ai.

Notes on sources

  1. Mistral AI, Mistral crawlers: MistralAI-Index, MistralAI-User and MistralAI-Training, user-agent strings and IP lists; the source of the quoted sentence on Mistral search. Undated page, later than 28 May 2026 because it names Vibe. Read on 10 November 2026.
  2. Mistral AI, Vibe gets to work, 28 May 2026; Mistral Help Center, Le Chat is now Vibe, updated on 12 August 2026. Read on 10 November 2026.
  3. Mistral AI, Search the web (Vibe Work): Web search, Open URL, Sources panel, AFP and AP; The all new le Chat, 6 February 2025. Read on 10 November 2026.
  4. Brave Software, About Brave, undated page; TechCrunch, article of 21 March 2025 on the February deal between Brave and Mistral. The original announcement was not found. Read on 10 November 2026.
  5. Mistral Help Center, Where do you store my data, updated on 12 August 2026. Read on 10 November 2026.
  6. Directive (EU) 2019/790, text on EUR-Lex, Article 4; United Kingdom, Copyright, Designs and Patents Act 1988, section 29A. The picture with the AI Act is in the guide to robots.txt for AI crawlers.
  7. The two itria.io cases come from the 25 September 2026 lab on Brave and Claude and from the study of food producers’ websites, both published on the site with their data.
·Who does it for you

Opening the door to Mistral is the first step. Getting chosen over your competitors is the work that follows. Positioning is built starting from the market.

Mattia Esposito, who wrote this guide, also works with Itria on positioning: in-depth market analysis to find where competitors are exposed and take that ground. He did it on his own site first. In August, asked who had founded itria.io, ChatGPT couldn’t find him. From 11 October to 7 November 2026 itria.io pages appeared 13,803 times in Google and Bing results and 3,818 times in AI answers: Google’s AI Overviews and AI Mode, and Microsoft Copilot.