GetLeads Contact us
← All articles

MegaTGAI · Practical guide

Telegram recommendations: a three-message example

A short conversation about a familiar business problem: a bartender taking sales off the books. See how the bar owner’s question leads to a recommendation and a reply from an AI staff-monitoring specialist.

Click a screenshot to enlarge it.

Feature examples

What it looks like in Telegram

Conversations, messages and profiles as other people see them.

Feature example

Question → recommendation → clarification

Alex says he has already fired a third bartender for off-the-books sales. Anna recommends Michael from the group, who helped link her cameras and till with AI. Michael recognises her and replies. MegaTGAI uses three selected accounts with different roles; this is a staged example.

The label colours match the message outlines. Click the phone for a closer look.

  • Question: Alex describes the problem and asks how to keep it under control.
  • Recommendation: Anna connects the solution to her problem with Michael and AI.
  • Clarification: Michael confirms he implemented the system and continues the conversation.

MegaTGAI setup

How to set it up in MegaTGAI

Follow the steps from selecting accounts to starting the task.

Step1

Choose suitable groups first

Choose groups relevant to your service.

Step2

Create a task and select three roles

Click ‘New task’ → ‘Recommendations’. One stream needs three different accounts: primary, advisor and confirmer. Select each role in turn and click ‘Done’. Then name and create the task.

Create a task and select three roles
  1. Recommendations mode and stream count.
  2. Open account selection.
  3. Tick the account for this role.
  4. Done — continue to the next role.
Step3

Add the selected groups

Open ‘Database’ on the task card. Select the task, paste your prepared group links and click ‘Add to task’.

Add the selected groups
  1. Check the selected task.
  2. Paste one link per line.
  3. Add the groups to this task.
Step4

Give each role its own AI instructions

Choose groups or channel comments. Select ‘AI instructions’ for each role, choose its preset and describe its part of the scenario. AI generates individual replies before sending. One pass is enough for a first run.

Give each role its own AI instructions
  1. Three separate roles with assigned accounts.
  2. All roles use AI instructions, each with its own preset and task.
Primary account

You are Alex, a bar owner. You have fired a third bartender for off-the-books sales. Ask other business owners how to control this. Keep it brief, without advertising.

Advisor

You are Anna. Reply to the question: Michael from this group helped connect your cameras and till using AI to identify discrepancies. Suggest asking him. Do not add new facts.

Confirmer

You are Michael, an AI specialist. Reply to Anna: you remember her bar, implemented staff monitoring and connected it to the till. Briefly confirm and continue without repeating her reply.

Prepared messages also support spintax: {Could you advise|Can anyone suggest}. It inserts one of the options inside braces.

Step5

Edit the selection and start

In the header, ‘Edit’ opens the accounts for that role. At the bottom, ‘Save changes’ only saves; ‘Save and start’ begins the task. The task card also offers ‘Settings’ and a separate ‘Start’ button.

Edit the selection and start
  1. Edit the account for a role.
  2. Save without starting.
  3. Save and start.
  4. Open task settings.
  5. Start from the task card.

After launch

Group and comment replies in one inbox

Replies to your messages, comments and recommendations arrive in the community inbox. Choose a task or ‘All new’, open a conversation and continue it right there, using the same account.

Group and comment replies in one inbox
  • Tasks and ‘All new’ collect replies from different communities.
  • ‘Needs a reply’ helps you find conversations awaiting a response.
  • On the right: the conversation and a reply field for the same account.

When direct-message intake is enabled, new inbound conversations appear under ‘Direct messages’.

The conversation, service, people and group size are staged examples, not published customer testimonials. The application capture uses isolated data and sends no messages.