Skip to main content

Building and Configuring an Agent in SimplAI Studio

SimplAI makes it easy to develop intelligent agents tailored to your specific needs.

Follow the steps below to create and configure an agent, equip it with the necessary tools, and test its responses.

  • Creating Agents: Define the agent's identity, capabilities, and tasks.
  • Configuring Agents: Assign models, knowledge bases, and tools.
  • Using Agents: Interact with agents through the built-in chat window.

Create a New Agent​

To create a new agent, follow these steps:

  1. Log In: Access your account at SimplAI.

  2. Go to Agents: Click on the Agents section in the sidebar.

  3. Create a New Agent: Click on + Create Agent at the top-right corner.

  4. Enter Agent Details: Provide a name and description for your agent.

    Screen Recording

Configuring Your Agent​

Once your agent is created, configure the following settings to ensure it functions as desired.

1. Choose the LLM Model​

  • Select the Large Language Model (LLM) that powers the agent.
  • Adjust Advanced Options (if required):
    • Max Tokens: Limit response length.
    • Top K Sampling: Define how many top tokens to consider.
    • Repeat Penalty: Prevent repetitive outputs.
    • Min P & Top P Sampling: Fine-tune token selection.

2. Define Base Instructions​

  • Set clear and structured instructions for how the agent should behave.
  • Example: "You are an AI customer support assistant. Answer queries politely and concisely."

3. Add a Knowledge Base​

  • Attach a Knowledge Base for accurate and structured responses.
  • Example: If building an UNO Rulebook Agent, add the UNO rules dataset.

4. Add Tools​

  • Equip the agent with necessary tools like Web Research, API calls, or PDF Parsing.
  • Tools allow agents to fetch external information or process files dynamically.

5. Set a Welcome Message​

  • Define a custom greeting to personalize user interactions.
  • Example: "Hello! How can I assist you today?"

6. Configure Guardrails​

  • Add Guardrails Validator to enforce response restrictions.

  • Useful for content moderation, data validation, or ethical AI guidelines.

    Screen Recording

Testing & Publishing Your Agent​

1. Preview and Test the Agent​

  • Navigate to the Preview Section.

  • Ask queries to the agent to test how it responds.

  • Example:

    • If using the UNO Rulebook Agent, ask: "What happens if I play a Wild Draw Four?"
    • If using the Web Research Tool, ask: "Find recent news about AI advancements."

    Screen Recording

2. Submit and Publish the Agent​

  • Once satisfied with testing, click Submit to save the version.

  • Click Publish to make the agent available for use in workflows.

    Screen Recording

Next Steps​

Now that your agent is live, you can:

  1. Integrate it into chat workflows or customer support tools.
  2. Enhance its capabilities with additional tools and knowledge bases.
  3. Monitor & Optimize performance based on user feedback.

By following these steps, you will have a fully functional AI Agent tailored to your specific requirements.