Tutorial

Learn MyBrandMetrics workflows

Step-by-step guides for connecting data, running agents, and using the workspace with fewer mystery buttons and more momentum.

Intro

Create an agent instance

Create a Hermes instance from the MyBrandMetrics Agents page.

Create an agent instance

Use this guide to create a Hermes agent instance from the MyBrandMetrics workspace.

Prerequisites

  • You are signed in to MyBrandMetrics.
  • You can open the Agents page from the left navigation.
  • You have at least one LLM provider available, or you want MyBrandMetrics to prepare one for you.
  • Optional: collect Telegram, WhatsApp, or Openviking settings before opening the form.

Step 1: Open Agents

From the left navigation, click Agents.

The page shows existing instances, status filters, pagination, and the Create instance button.

If you do not see a recent instance, click Refresh List.

Step 2: Open Create instance

Click Create instance.

The modal submits to the MyBrandMetrics agent proxy and uses the same payload shape as the backend instance API.

Fill the required top-level fields:

  • Name: a friendly instance name, for example marketing-agent.
  • Instance type: choose Hermes.

Step 3: Configure Providers

The Providers section controls built-in model providers.

Each provider has:

  • Provider name
  • Model ID
  • API key
  • Default selection

Choose MyBrandMetrics if you want the workspace-managed provider path.

When Provider name is MyBrandMetrics:

  • Model ID can be selected or typed manually.
  • The default model is minimax-m3.
  • Available model options include minimax-m3, minimax-m2.7, gpt-oss-120b, deepseek-v4-flash, and deepseek-v4-pro.
  • If your account does not have a permanent API key, MyBrandMetrics creates one automatically and fills it in.

You can also choose another provider such as Google, OpenAI, Anthropic, Grok, Qwen, Kimi Moonshot, OpenRouter, MiniMax, Deepseek, or Xiaomi.

When switching provider names, the API key field resets. This avoids accidentally sending one provider's secret to a different provider.

Click Add provider if you want multiple built-in providers, then use Set as default to choose which one the instance should prefer.

Step 4: Add Custom Providers if needed

Use Custom providers only when you have your own compatible endpoint.

Each custom provider requires:

  • Provider name
  • Model ID
  • Base URL
  • Endpoint compatibility
  • API key

The default endpoint compatibility value is usually:

openai-completions

You can mark one custom provider as default, but only one provider across built-in and custom providers should be the default.

Step 5: Configure optional integrations

The form includes optional collapsible sections in this order:

  • Telegram
  • WhatsApp
  • Openviking

Telegram

Open Telegram if you want the instance to respond through a Telegram bot.

Fill:

  • Telegram bot token
  • One or more Telegram user IDs

If you fill a bot token without user IDs, the form will ask you to add at least one user ID. If you add user IDs without a bot token, the form will ask for the bot token.

WhatsApp

Open WhatsApp if you want to allow specific phone numbers.

For each number:

  • Choose the country code.
  • Enter the phone number digits.

The form validates phone numbers and supports multiple allowed numbers.

Openviking

Open Openviking only if you need Openviking model settings.

Optional fields include:

  • Embedding model
  • Embedding API key
  • VLM model
  • VLM API key

Step 6: Create the instance

Click Create instance.

While the request is running, the button changes to Creating....

After creation succeeds:

  • The modal closes.
  • The instance list refreshes.
  • The new instance appears with a status such as queued, starting, or running.

Use Refresh Status or Refresh List while provisioning finishes.

Step 7: Open or manage the instance

When the instance is running, the card can show:

  • Access URL
  • Gateway token or password details when available
  • Open terminal
  • File Manager
  • Start, stop, delete, and refresh actions depending on status

For Hermes instances, the displayed Access URL is the base Clawbus URL only, without the /#token=... fragment.

Verification checklist

  • The instance appears on the Agents page.
  • Status eventually becomes running.
  • The Access URL opens successfully.
  • If you configured Telegram or WhatsApp, send a test message from an allowed account.
  • If you need server access, use Open terminal or File Manager from the instance actions.

Troubleshooting

Create instance fails

  • Confirm Name is filled.
  • Confirm at least one provider has Provider name, Model ID, and API key.
  • If using MyBrandMetrics, wait for the permanent API key to be prepared.
  • If using a custom provider, confirm Base URL and Endpoint compatibility are correct.

Telegram validation fails

  • Fill both Telegram bot token and at least one numeric Telegram user ID.
  • Remove spaces or non-digit characters from user IDs.

WhatsApp validation fails

  • Select the correct country code.
  • Enter only phone number digits.
  • Keep the full number length between 6 and 15 digits including the country code.

Instance stays in starting or queued

  • Click Refresh Status.
  • Wait a few minutes for provisioning.
  • If it remains stuck, stop/delete and recreate the instance after confirming provider credentials.