Privacy Policy
Crutor is an AI assistant that screens job candidates in Telegram, with a web app for the hiring team. This policy explains what Crutor stores, why, who sees it, how long it is kept and how to have it deleted. It covers the bot @Crutorbot, the web app and the site crutor.app.
The short version: we store what the service needs to run, we show it only to the hiring team it was sent to, we don't sell it, don't use it for advertising and don't train AI models on it, and we delete it when asked.
Who we are and how to reach us
Crutor is an independent project run by its developers. For anything about your data, write to support@crutor.app or to the bot @Crutorbot in Telegram. A person reads these messages and replies.
Who this policy covers
- Recruiters: people who create jobs and read applications.
- Team members: people a recruiter adds to a job's Telegram group.
- Candidates: people who apply through a job link, or who write to a recruiter's Telegram Business account that the recruiter connected to Crutor.
Candidate data is collected on the recruiter's behalf. The recruiter decides what to ask and what to do with the answers; Crutor runs the conversation on their instructions. Questions about why a particular company asks for your data go to that company; questions about how Crutor handles it go to us.
What we store
- Telegram account data: your Telegram id, name, username and language, as Telegram gives them to bots. In Business mode, also the connection id and the chats you route to the bot.
- Messages: everything sent to the bot, including documents, photos and voice messages, and in Business mode the messages in the connected chats in both directions. Each message is kept with its Telegram metadata: message id, dates, edits, forwarding origin. An edit keeps the previous version. Where Telegram reports a deletion, the copy is marked deleted and stays.
- Applications: answers to screening questions, the documents a candidate sends (CVs and the like) and the text extracted from them, summaries and fit scores written by the AI, and the hiring team's notes and statuses.
- Recruiter content: jobs and their questions, team contacts (names, roles, emails, aliases), tone-of-voice settings, and the dialogs a recruiter forwards or uploads to teach the AI their tone. Before a working copy is stored, links, emails, phone numbers and other people's names are removed from it; the forwarded originals stay in the message archive like any other message.
- Calendars: see the section below.
- Web and technical data: a session cookie after you log in, login requests, IP addresses and browser details in server logs, API tokens you create, and product events (which features are used). For analytics see the section below. There are no advertising trackers.
- AI records: the input and output of every AI call, with its cost, attributed to the recruiter it ran for. They are used to show usage, to debug and to investigate abuse.
Why
- To run the service: talk to candidates, build applications, book interviews, show the hiring team what came in.
- To show recruiters their AI usage and its cost.
- To keep the service safe: look into complaints, spam, abuse and attempts to manipulate the AI; enforce the Terms; restrict accounts that break them.
- To fix bugs.
- To answer requests from Telegram or from authorities where we are required to.
We don't use your data for advertising, we don't sell it, and we don't use it to train AI models.
AI processing
Conversations with candidates, documents, job texts and tone settings are processed by large language models run by third-party providers. We use them on commercial terms under which the provider does not use the data to train its models and does not keep it beyond what is needed to return a response. The AI suggests; the hiring team decides.
Who sees the data
- The hiring team of that job: the recruiter and the members of the job's group see the candidate's conversation, documents, answers, summary and score in the group and in the app.
- Candidates see their own answers and can correct them. They never see scores, notes, other candidates or the team's discussion.
- Crutor's administrators: only to operate the service, to support users and to look into abuse or a complaint. Every time an administrator opens a conversation, it is logged.
- Service providers: the hosting we rent, the AI providers above, the calendar providers you connect, PostHog for analytics (see below), and Telegram itself, through which every message passes.
- Tools you connect: a personal access token or an MCP connection gives the tool exactly what the token allows. That is the recruiter's choice, and a token can be revoked at any time.
- Authorities and Telegram: where the law requires it, or when we report abuse to Telegram and point to the messages concerned.
Nobody else.
Calendars
Scheduling is optional. A recruiter, a team member or a candidate can connect a calendar so that Crutor can offer interview times that are free for everyone. Access tokens and calendar addresses are encrypted at rest with a key kept apart from the database. Disconnecting a calendar deletes its token at once; a candidate's token is deleted as soon as the meeting is booked.
Google Calendar. Crutor asks Google for two permissions and uses each one for exactly one thing:
- See your free/busy status (
calendar.freebusy): to find out when you are busy in the time range being scheduled. Crutor never reads event titles, descriptions, attendees or locations. - Edit events on calendars you own (
calendar.events.owned): to create the interview events that Crutor books on your own calendar, with a Google Meet link, and to move or cancel them later. Crutor does not read, change or delete any other event, and never touches calendars shared with you.
Free/busy answers are used to compute the slots to offer and are not kept as a copy of your calendar; what we keep is the slots we offered and the meetings we booked. Google data is shared only as the feature requires: an offered time with the candidate, the meeting link with its participants. It is not used for advertising, not sold, and not used to train AI models. Crutor's use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. You can disconnect in the app under Settings → Calendars or with /calendar in the bot, and revoke access at any time in your Google account.
Calendar links (ICS). If you paste a calendar address instead, it is stored encrypted and read only for busy intervals.
Analytics
To understand how people use the site and the bot, we use PostHog, an analytics service hosted in the EU (Frankfurt).
- On the site (the landing, the guide and the legal pages; not the web app): page views, clicks and, for a share of visits, a replay of the page as you saw it (scrolling, mouse movements, what was shown). Everything typed into a field is masked in your browser before anything is sent. PostHog estimates your country and city from your IP address.
- No cookies: the anonymous visitor id is kept in the browser tab's session storage, so the pages you open in one tab count as one visit, and it is deleted when you close the tab. Nothing is kept in cookies or in your browser's permanent storage. Requests go through our own domain. If your browser sends Do Not Track or Global Privacy Control, the analytics are not loaded at all.
- From the site to the bot: once the analytics have loaded, the site's links to the bot carry that anonymous id, and when you start the bot from one we link the visit to your Crutor account under an internal id. Your Telegram id, name and messages are not sent.
- In the bot and the app: product events, such as tour steps, choosing to hire or to look for a job, and a recruiter's first job created and published, are sent under the same internal id. Message texts and data about candidates are never sent.
Where the data is and how it is protected
Data is stored in our own database and file storage on servers we rent from a cloud hosting provider. Connections to the site and the app are encrypted; messages travel through Telegram's own transport. Tokens and calendar addresses are encrypted with a key kept separately. Documents can be downloaded only through short-lived signed links. Administrators' access to conversations is logged. If a security incident affects your data, we will tell you as the law requires.
How long we keep it, and deletion
We keep data for as long as the service needs it, and we do not delete messages and documents on a timer: they are kept so that a complaint or an abuse report can be looked into after the fact.
To have your data deleted, write to support@crutor.app or to the bot @Crutorbot from the Telegram account the data belongs to. A recruiter can also ask on behalf of their candidates. We delete without undue delay, except what we have to keep to look into an open abuse case, to defend a claim, or to meet a legal obligation; that is deleted once the reason is gone. Deletion is done by a person and recorded in our audit log.
Withdrawing an application marks it as withdrawn but keeps the conversation; ask us if you want it gone. If Crutor shuts down, the data is deleted.
Your choices
- See what we hold about you and get a copy: write to us.
- Correct it: candidates can change their answers in the recap the bot sends; recruiters edit everything in the app.
- Delete it: see above.
- Disconnect a calendar, revoke an API token, log out of browser sessions (/sessions in the bot), withdraw an application, or stop the bot: all available at any time.
Whatever the law where you live calls these rights, write to us and we will act within a month.
Children
Crutor is for people of working age. We do not knowingly store data about anyone under 16, and recruiters must not use Crutor for jobs aimed at children. If you believe we hold such data, tell us and we will delete it.
Changes to this policy
Changes are published on this page with the date. If a change affects what we store or who sees it, the bot will tell you before it takes effect.
Languages
This policy is published in English, Russian and Ukrainian. If the translations differ, the English text applies.