How to create and use assistants

Go to the Admin Panel, located in the top right corner of the screen.

Go to the "Assistants" section in the menu on the left side of the screen.

Click "New assistant."

Fill in the assistant’s details.

• Name the assistant
  • In the "Name" field, enter a name that reflects the assistant's purpose.
• Provide a description
  • In the "Description" field, write a short description that explains the main function or goal of the assistant. This helps understand when to choose this assistant.
• Set the assistant icon (optional)
  • If you want to customize the assistant’s icon, you have two options
  • Regenerate - Click to automatically generate an icon.
  • Upload Image - Alternatively, you can upload a .png or .jpg file to set your own image for the assistant. This will override any generated icon.
• Write the system prompt:
  • In the "System Prompt" section, provide a detailed description of what the assistant should do. This may include the target audience, main goal, and specific tasks it should assist with.
  • Example - "You are an assistant for on-call engineers. Your goal is to review the provided contextual documents and give recommendations on how to solve the problem."
• Choose the LLM provider:
  • From the "LLM Provider" dropdown menu, select the language model that will power the assistant.
• Enable advanced features (optional):
  • If you need advanced capabilities, select the appropriate options
  • Image Generation Tool - Allows the assistant to create images based on text descriptions.
  • Search Tool - Enables the assistant to search for information in real-time.
• Add additional instructions (optional):
  • In the "Additional Instructions" field, provide specific directions to guide the assistant’s actions when handling user queries.
• Add starting messages (optional):
  • If you want the assistant to start conversations with predefined messages, click the green ADD NEW button. This allows setting an initial message users will see at the beginning of interactions with the assistant.
  
• Set availability:
  • Is Public - Check this box if you want the assistant to be available to all users. If left unchecked, only specific user groups will have access to the assistant.
• Finish and create:
  • After configuring all options, click the CREATE button at the bottom to save and launch the assistant.

In the "Assistants" tab, found in the left menu, you will be able to see all your assistants and modify them as needed.

Go to "AI chat" at the top of the screen.

You will be able to select the assistant you want to use from the dropdown list.

After choosing the assistant, you can ask a question. The assistant, using connected data and its instructions, will provide an answer.

Sources for the answer can be viewed on the right side of the screen and as links directly in the response.

• Set clear and specific goals
  • When configuring the assistant, carefully consider its main goal. The more specific the goal, the better the assistant will perform. A clear goal also helps set the system prompt and choose the right LLM, ensuring relevant answers.
• Customize system prompts for precision
  • Adjust the system prompt to give the assistant specific guidance about answers, tone, and knowledge areas. This can include considering typical queries, preferred answer styles, or industry-specific terminology.
• Regularly update and refine the assistant’s scope and content
  • As business needs evolve, the assistant should evolve too. Regularly reviewing and updating the system prompt, knowledge base, and features ensures it remains current and useful.
• Formulate questions clearly and concisely
  • Clarity and conciseness are key for effective prompt use. When questions are clear and specific, the assistant better understands intent and provides relevant answers. Avoid overly complex language and ambiguous terms, focusing on essential details. For example, instead of asking "Tell me everything about data science," use "Give a brief overview of data science and its main areas."