Multi-Bot Telegram Customer Service System Construction Guide: Project Isolation and Agent Cross-Project Reception Practice

When operating a single-brand Telegram Bot, the customer service process is relatively simple: one Bot, one backend, and one team. But things get complicated when your business expands to multiple brands, product lines, or communities. Each Bot requires independent customer service configuration, but the team hopes to manage it uniformly in one backend. This is the core problem that the multi-bot Telegram customer service system needs to solve.

This article will take TG-Staff as an example to teach you step by step how to build a multi-bot customer service system that supports project isolation and agent cross-project reception. Whether you are a cross-border e-commerce company operating multiple store groups, a Web3 project maintaining multiple token Bots at the same time, or a SaaS company configuring independent customer service for different product lines, this solution can help you achieve a balance between centralized management and flexible deployment.

Why does multi-brand operation require an independent Telegram customer service project?

Before starting to build, first understand “why project isolation is done”. Many teams initially used only one Bot to handle all user inquiries. However, as the business expanded, problems gradually emerged:

• Brand Confusion: User A inquired about products of brand
• Data Mixed: All sessions, user portraits, and statistics are mixed together, making it impossible to analyze independently by brand.
• Permissions confusion: The intern agent was supposed to only handle simple inquiries from Brand Z, but was able to see sensitive conversations from Brand X.

Project Isolation is designed to solve these problems - each Bot project has independent configuration, session records, user data and risk control rules. But at the same time, the operation team often hopes that customer service staff can log in to a web backend and receive multiple Bot users at the same time, instead of opening a separate backend window for each Bot. This leads to the need for agent sharing.

Common multi-Bot scenarios: cross-border multi-stores, Web3 multi-chains, SaaS multi-products

• Cross-border e-commerce: Operates multiple Telegram groups in the US, Europe, and Japan. Each group corresponds to a Bot. The customer service team uses Chinese + automatic translation to receive messages.
• Web3 Project: Maintain mainnet token Bots, testnet Bots, and NFT project Bots at the same time. Each Bot has different community rules and risk control requirements (such as monitoring wallet addresses of different chains).
• SaaS company: Configure independent Bots for CRM products, marketing tools, and data analysis tools, but the customer service team shares a set of knowledge base and agent accounts.

Project isolation vs. agent sharing: How to get the best of both worlds?

To put it simply: Project isolation is independence at the data level, and Agent sharing is unification at the operation level. The two are not contradictory.

• The session data of Project A can only be seen by the administrator and designated agents of Project A.
• Agent Xiao Zhang can join Project A and Project B at the same time, and switch to handle the consultations of the two Bots in the same web background.
• The conversation diversion rules (such as “online priority”), automatic translation configuration, and content risk control phrases of Project A will not affect Project B.

TG-Staff is based on this design concept - each Bot project is an independent “container”, and agents can work across containers through permission binding.

Step one: Create multiple Bot projects in TG-Staff

The steps are very straightforward and you don’t need development experience.

1. Log in to the console: Visit https://app.tg-staff.com/ and complete the registration (free trial for 3 days).
2. Add the first Bot project: Click “New Project” and enter the project name (such as “Brand X Customer Service”).
3. Connect Telegram Bot: Create your Bot in @BotFather, copy the API Token, and paste it into the TG-Staff project configuration page.
4. Edit Bot Information (optional): Modify the Bot’s avatar, name and description directly in the console without jumping to BotFather. This feature is useful for quickly adjusting your brand image.
5. Repeat the above steps: Create independent projects for the second and third Bot.

Once created, you will see a list of all projects in the console. Each project has an independent settings page, including configurations such as session offloading, automatic translation, and content risk control (Professional version).

Step 2: Configure project customer service scope and agent permissions

The flexibility of the multi-Bot system is reflected here: you can specify “who can answer the customer service of this Bot” for each project.

Set project-level customer service scope: all customer service vs. designated customer service

Enter project settings → “Customer Service Scope” and you will see two options:

Mode	Applicable Scenarios	Description
All Customer Service	Multi-Bot shared customer service pool	All agents with permissions for this project can receive users of this Bot. Suitable for general consultation.
Designated Customer Service	Exclusive Agent/Advanced Support	Only the specific agent you checked can receive you. Suitable for Bots that require professional knowledge (such as technical after-sales service).

Best Practice: For standardized pre-sales consultation bots, it is recommended to use the “all customer service” mode to maximize agent utilization. For Bots that require in-depth product knowledge (such as enterprise-level SaaS troubleshooting), use the “designated customer service” mode to avoid incorrect responses from non-professional agents.

Assign cross-project permissions to agents

On the agent management page, you can assign an agent account to multiple projects at the same time:

1. Invite or add an agent (enter the other person’s Telegram account or email).
2. On the agent details page, click “Assign Project”.
3. Check the items that the agent needs to serve (such as “Brand X Customer Service” and “Brand Y Customer Service”).
4. Set a separate role for each project: Administrator (can modify project configuration) or Agent (can only receive sessions).

In this way, after agent Xiao Zhang logs in to the backend, he will see all pending sessions from project A and project B. He doesn’t have to switch accounts or open multiple browser tabs.

Step 3: Use session diversion and diversion links to achieve multi-channel traffic drainage

After the project is built and the agents are assigned, the next thing to consider is: how do users find the corresponding Bot from different channels? And when the number of consultations increases, how to automatically assign them to online agents?

Configure session diversion rules

Each project can set up diversion rules independently:

• Rotating Assignment (Default): New user sessions are assigned to authorized agents in sequence. It is suitable for scenarios where the number of agents is stable and the workload is balanced.
• Online Priority: Prioritize allocation to currently online agents; if all agents are offline, fallback to turn-based allocation. Suitable for teams with a schedule system or agents whose online hours are not fixed.

Use diversion links (magic links) for channel attribution

TG-Staff’s Diversion Link (officially called Diversion Link) is a powerful traffic attribution tool. It is a short link to the official domain name of TG-Staff (such as https://app.tg-staff.com/{code}). After users click it, they will automatically jump to your Telegram Bot and capture the following data before jumping:

• Visitor IP address
• Browser information (User-Agent)
• URL parameters (you can customize utm_source, utm_medium, etc.)

Actual usage:

1. Generate a diversion link for Brand X’s Bot for Google Ads advertising.
2. Generate another diversion link for Brand X’s Bot for Twitter promotion.
3. Generate independent diversion links for Brand Y’s Bot for Facebook ads.

After the user clicks on different links to enter the Bot, TG-Staff will record the source channel in the background. In the subsequent user portraits and statistics, you can analyze which channel brings the largest amount of inquiries and the highest conversion rate.

Step 4: Actual Agent Practice—Cross-Project Reception and Collaboration

When everything is configured and agents log in to the TG-Staff Web backend, they will see a unified session list. Here are a few key operating tips.

How to quickly identify the project to which the current session belongs

The backend UI has a clear design:

• Project Name and Bot Name will be displayed at the top or side of each conversation.
• The session list supports filtering by project, and agents can only view the sessions of a certain Bot.
• If an agent is working on multiple projects at the same time, it is recommended to use the “Group by Project” view to avoid confusion.

Common Mistake: The agent mistakenly sends the words of Project A to the user of Project B. TG-Staff’s UI design reduces this risk through color labels and item identification, but the habit of “see items first, then reply” should still be emphasized in agent training.

Cross-project session transfer and collaboration skills

When encountering a problem that cannot be solved by oneself, agents can use the Session Transfer function:

• Transfer the session to another agent (such as a technical expert) working on the same project.
• Before transfer, you can add Private Note (Professional version feature) to record the background of the problem and the steps that have been tried to help the taking over agent quickly understand the context.

Collaboration scenario example:

• Agent Xiao Zhang is receiving a user from brand X, and the user asked an API integration question. Xiao Zhang is not a technician, so he adds a note “The user has provided API logs and requires technical troubleshooting”, and then transfers the session to Xiao Li, a technical agent under the same project.
• After Xiao Li took over, he saw the content of the note and checked the log directly without asking the user repeatedly.

FAQ

**Q: How many Bot projects can a TG-Staff account manage at most? **

A: Depends on the package selected. Both the standard version and the professional version support multi-project management. For the specific number of projects, please refer to the official website package page. You can add or delete Bot projects at any time in the console, and there is no fixed “minimum number of projects” limit.

**Q: Can the same agent receive multiple Bot users at the same time? Will it be confusing? **

Answer: Yes. After logging in to the TG-Staff Web backend, agents will see a list of sessions from different Bots, and each session is clearly marked with the name of the project it belongs to and the name of the Bot. The system also supports filtering sessions by project, so agents can focus on a certain Bot’s queue. It is recommended that agents develop the habit of “looking at the item tag” before responding.

**Question: If the consultation volume of a certain Bot suddenly increases sharply, how to quickly deploy other agents to support it? **

Answer: There are two ways: 1) Change the “Customer Service Scope” from “Specified Customer Service” to “All Customer Service” in the project settings so that all authorized agents can access; 2) Manually add more agents to the project. At the same time, change the offloading rule to “online priority”, and the system will automatically assign new sessions to currently online agents to maximize response speed.

**Q: Can the diversion link (magic link) be used for multiple Bots? **

Answer: Yes. Each Bot project can generate multiple independent diversion links. You can generate and manage links for different Bots and different advertising channels separately, and the attribution data will not interfere with each other. For example, the Google Ads link for Brand X and the Twitter link for Brand Y will record their respective click and session data in the background.

**Q: Can content risk control (internal control management) set different rules for different projects? **

Answer: Yes (Professional version function). You can create multiple risk phrases in the console, and then associate different phrases to different projects. For example, Project A (Mainnet Bot) monitors TRC20 wallet addresses, Project B (Testnet Bot) monitors ERC20 addresses, and the rules are completely independent. When an agent sends a message in Project A, only the risk word detection associated with Project A will be triggered.

Conclusion: Unified backend, flexible service for multiple brands

The core value of building a multi-bot Telegram customer service system is: centralized management reduces costs, project isolation ensures data security, and flexible agent deployment improves efficiency. No matter your business spans several brands or product lines, a properly configured TG-Staff backend can make it easy for you.

If you are having trouble managing customer service for multiple Bots, you might as well start with a free trial:

1. Visit https://app.tg-staff.com/ to register and enjoy a 3-day free trial.
2. Create the first Bot project in the console and experience the multi-project management process.
3. Check Official Documentation for more detailed configuration guide.
4. If you encounter any problems, please contact the customer service Bot @tgstaff_robot directly.

You are only one step away from an efficient multi-bot customer service system.