Telegram Bots for Internal Processes: HR, Approvals and Reports
Basics4 minInnovialabs editorial team
In short
An internal Telegram bot handles short, repetitive actions well — requests, approvals, reports and alerts from business systems — but data and access rights must live in your systems, not in the chat.
The short answer: why a company needs an internal bot
Employees already open Telegram dozens of times a day. If a short work action can be done right there — requesting leave, tapping "Approve", submitting an end-of-shift report — people do it faster and more often than through a corporate portal that needs a separate login.
An internal bot is not a replacement for your business system, it is a convenient entry point into it. Data stays in 1C, the CRM or the HR system; the bot only accepts actions and delivers notifications.
Typical use cases
- Onboarding. A new hire gets a link to the bot and walks through a first-week plan: documents, mentor contacts, policies, a short quiz. The manager sees which step the person is on.
- Leave and time-off requests. The employee picks dates with buttons, the bot checks remaining days in the HR system and sends the request to the manager.
- Approvals. An invoice, contract or purchase arrives as a message with "Approve" and "Reject" buttons and a comment field. The decision is written back to the system.
- Daily reports. At the end of a shift the bot asks 3–5 questions (revenue, stock, issues) and collects the answers into a spreadsheet or BI tool.
- Alerts from 1C and CRM. A large new deal, an overdue payment, negative stock, a server going down — the system itself messages the right person or work group.
Access control
This is the key question for an internal bot. Bots are public: anyone can find one by its username and press Start.
- Link the account to an employee. On first launch the bot asks for a phone number via a "Share contact" button, or a one-time code sent to the corporate email. The number is matched against the employee directory.
- Store the link. The bot stores the Telegram
user_id(not the username, which can change) and the employee's role. - Check on every action. Permissions are checked on the server on every button press, not only at login. A button in an old message must stop working once a person leaves or changes position.
- Revoke on offboarding. Deactivation in the HR system automatically blocks bot access and removes the person from work groups.
- Keep sensitive data out of the chat. Salaries, personal data and commercial terms are better shown via a link to a protected page than as plain text in the conversation.
When a bot beats a corporate portal
| Task | Bot | Portal |
|---|---|---|
| One action in 1–3 steps | Easier | Overkill |
| Reaction needed within minutes | Push arrives instantly | Only when the tab is opened |
| Field, warehouse or retail staff | Works from a phone | Often awkward |
| Complex forms, tables, filtered reports | Cramped | Better |
| Working with documents and history | Links only | Main tool |
A good setup is bot plus web interface: short actions and alerts live in the bot, while complex screens open as a Mini App or a web page from a button.
Limits of Telegram as a work tool
- Personal accounts. Employees use their own numbers, and the account belongs to them, not to the company. You cannot "take back" a private chat with the bot when someone leaves, so anything important must be stored in your systems.
- Data requirements. Some industries and types of personal data may come with rules about where data is stored and processed. Check this with a lawyer before launch.
- Bot API limits. Mass notifications to thousands of employees must respect message rate limits.
- Availability. If Telegram is unavailable on the company network or for some employees, critical processes need a fallback channel.
- Noise. If the bot sends everything, people start ignoring it. Only send what requires action.
Common mistakes
- Letting in anyone who knows the bot's username without checking the directory.
- Keeping roles and permissions only inside the bot instead of the business system.
- Building long questionnaires with dozens of questions in chat instead of a form.
- Not logging who pressed "Approve" and when — disputes need an audit trail.