Building a Personal AI Assistant with OpenClaw (Telegram)

Scenarios

With the advancement of AI technologies, personal AI assistants have become essential tools for improving work efficiency and quality of life. In daily online communication, people often need to interact with various AI assistants to obtain information or complete tasks. However, these AI assistants are usually limited to specific platforms or models, and have certain usage restrictions. As a result, users have to frequently switch between different platforms. This increases operation complexity.

OpenClaw (formerly known as Moltbot and clawdbot) is a continuously running AI agent that can invoke multiple large language models. Acting as a gateway, OpenClaw lets you interact with multiple AI models through messaging channels. This eliminates the need to switch between different platforms and streamlines the operation process.

This section describes how to deploy the OpenClaw application image on a FlexusL instance and integrate it with Telegram to quickly build a personal AI assistant.

The OpenClaw application image provided by FlexusL is installed using official open-source scripts. Before deploying the image, read and acknowledge the risks described in OpenClaw Risk Notice and Security Suggestions.

The following steps describe how to purchase a FlexusL instance and configure a messaging channel from scratch. If some of the steps have been completed, you can skip them.

Step 1: Deploy a FlexusL Instance with the OpenClaw Application Image

You can deploy the OpenClaw application image on a new or an existing FlexusL instance.

Step 2: Create a Telegram Bot

1. Sign in to Telegram, enter @BotFather in the search box, and click the displayed contact.

   

2. In the dialog box, enter /newbot.
3. Enter the bot name and username as prompted. The username must end with _bot.
4. After entering the information, you will receive a token. Keep it secure for subsequent OpenClaw configuration.

   

Step 3: Configure the OpenClaw Application

1. Log in to the FlexusL console.
2. Click Configure Now on the target FlexusL instance card. The Application Management page is displayed.

   If Configure Now is not displayed, click the resource card to go to the resource details page and choose Application Management from the left navigation pane.

   Some functions of the OpenClaw application image (such as querying gateway status and updating models and application configurations) depend on Cloud Operations Center (COC). You need to enable and authorize COC before using these functions. If the content in the following figure is displayed, enable and authorize COC as prompted.

   

3. In the Basic Information area, check that the UniAgent status and gateway status are running.
   

   Table 1 Basic application information

   Parameter	Example	Description
   UniAgent Status	Running	Operations such as querying the OpenClaw gateway status and delivering configuration scripts depend on the unified data collection agent (UniAgent) provided by Application Operations Management (AOM). After you purchase a FlexusL instance, the UniAgent is installed automatically. The UniAgent status is as follows: Running: The UniAgent is working. Installing: The UniAgent is being installed. Wait until the installation is complete and then perform operations on the FlexusL instance. Abnormal: The UniAgent is abnormal. Installation failed: The UniAgent fails to be installed. Not installed: The UniAgent has not been installed. NOTE: If the UniAgent status is Abnormal, Installation failed, or Not installed, manually install the UniAgent by referring to Installing UniAgents.
   Gateway Status	Running	Status of the OpenClaw gateway. Running: The OpenClaw service is working normally. Unknown/Unhealthy: The initial status of the OpenClaw gateway is Loading. If the status changes to Unknown or Unhealthy after several minutes, click Restart. The restart takes some time. After the gateway is restarted successfully, the status changes to Running. NOTE: If the gateway status is still Unknown or Unhealthy after you click Restart, submit a service ticket.
   OpenClaw Version	2026.3.24	OpenClaw version. For more version details, see Upgrading the OpenClaw Version.

4. In the model configuration area, configure models and channels and install skills as needed.

   Configure Models

   1. Click Configure Models. Then, click Add Model and configure model details in the displayed right pane.

      

      Table 2 Model configuration parameters

      Parameter	Example	Description
      Model Provider	Custom	Select Custom and enter required model parameters in JSON format in the text box. base_url: The API address of the model. Custom models only support OpenAI-compatible API addresses, for example, https://api.openai.com/v1. Obtain the OpenAI-compatible API address from the model provider. api_key: The API key of the model provider. model_id: The unique identifier of the model.
      Model Alias	ds3.2	Enter a custom model name that is easy to identify.

   2. After the model is configured, click OK. You can view it in the model list. The first added model is automatically set as the default model.
   3. Add more models as needed. You can click Set as Default in the Operation column to change the default model.

      The new default model will not be used unless you start a new session. You can enter "/new model-alias" (start a new session) or "/model model-alias" (retain the context) in the chat box.

      

      When you enter /new in the chat box to start a new session, the bot will lose its memory. Exercise caution when performing this operation.

   4. Delete models as needed.
      

      After a model is deleted, the bot cannot use it anymore. To reuse the same model, add it again.

   Configure Channels

   1. Click Configure Channels. Then, click Add Bot and configure bot details in the displayed pane.
      

      Table 3 Bot configuration parameters

      Parameter	Example	Description
      Channel	Telegram	Select the channel to be connected.
      Bot details	bot1 Bxxxxxxtoken	Bot alias: Enter a custom alias that is easy to identify. Token: Enter the token of the bot created in step 2.

   2. After the bot is configured, view it in the bot list.
      

      • You can add more bots as needed. A maximum of 10 bots are allowed for all channels.
      • If you delete channels or bots, the configured bots will no longer be available. Exercise caution when performing this operation.

   Configure Skills (Optional)

   1. Click Configure Skills (Optional). The installed skills are displayed in the skill list. For details about the skills, visit ClawHub Community Skills.

      

   2. Click Add Skill. In the displayed pane, search for the skills to be installed by name. Currently, only official skills can be installed. For details, see ClawHub Community.

      

   3. After the installation is successful, view the installed skills in the skill list.

      

      

      If a message is displayed indicating that the ClawHub community is busy and the installation fails, perform the operations described in What Should I Do If Skills Failed to Be Installed?

(Optional) Step 4: Configure Security Groups

When you create a FlexusL instance, the system automatically creates a default security group (sg-default-smb) and associates it with the instance. The default security group only allows instances in the security group to communicate with each other but denies all external requests by default. In this example, use the default security group and default security group rules. You can perform the following operations to view and configure the security group rules.

1. Log in to the FlexusL console and click the target resource card or instance name to go to the instance details page.
2. In the navigation pane on the left, choose Cloud Servers and click the target server name to view its details.

   

3. Click the Security Groups tab and check that the default security group rules are used.

   You can also add more rules as needed. For details, see Configuring Security Group Rules for a FlexusL Instance.

   

   If the source of an inbound rule is set to 0.0.0.0/0 or ::/0, then all external IP addresses are either allowed or denied to access your instances, depending on if the action is Allow or Deny. If the access is allowed, exposing high-risk ports, such as port 22, 3389, or 8848, to the public network will leave your instances vulnerable to network intrusions, service interruptions, data leaks, or ransomware attacks. You should only configure trusted IP addresses for the source in security group rules.

   

   Table 4 Security group rules

   Priority	Action	Type	Protocol & Port	Source	Description
   1	Allow	IPv4	TCP: All	sg-default-smb	Allows instances in the security group to communicate with each other over IPv4 protocols.
   1	Allow	IPv6	TCP: All	sg-default-smb	Allows instances in the security group to communicate with each other over IPv6 protocols.

Step 5: Verify the Bot Conversation

1. Return to the chat box of BotFather in step 2 and click the provided link to start a conversation with your bot.

   

2. Click START.

   

3. If you can communicate with the bot, the connection is successful.

   

(Optional) Step 6: Reset the Password and Log In to the FlexusL Instance

A FlexusL instance does not have an initial password. After creating a FlexusL instance, you need to set a password before logging in to it. For details, see Resetting the Password for a FlexusL Instance.

1. Locate the resource card and choose > Reset Password. Alternatively, in the list view, click Reset Password in the Operation column to reset the password for the FlexusL instance.
   Figure 1 Resetting a password
   
2. Locate the resource card and click Remote Login. Alternatively, in the list view, click Remote Login in the Operation column. In the displayed dialog box, click VNC Login and enter the username and password as prompted to log in to the FlexusL instance. For more login methods, see Login Modes.

   The username is root, and the password is the one you set in the previous step.

   

Helpful Links

• To configure or modify the connected AI models or channels, see Configuring the OpenClaw Application.
• You are advised to select 2 vCPUs | 4 GiB of memory or higher to run the OpenClaw application image. To upgrade specifications, see Modifying the Specifications of a FlexusL Instance.
• For OpenClaw FAQs, see OpenClaw FAQs.