Skip to main content

Getting Started

Creating Your First Simple Agent & Task

Learn to create your first simple AI agent in Portal One! Step-by-step guide to build a text summarizer, connect an LLM, and get your first AI response via chat.

Welcome! This guide will walk you through creating your very first AI agent in Portal One and giving it a simple task. This will help you understand the basic workflow.

Our Goal: Create an agent named "Summarizer" that can summarize a piece of text you provide in a chat conversation.

Before You Start:

Let's get started!

Step 1: Create a New Agent

  1. From the main Agents view (where you land after login or by navigating to "Agents"), find and click the "+ New Agent" button.
    1. This opens the "New Agent" dialog.

Step 2: Define Your Agent

In the "New Agent" dialog, provide some basic details for your agent.

  1. Name: Summarizer
  2. Description: An AI assistant that helps summarize text.
    1. (Tip: The Name and Description help your agent understand its role and establish its identity.)
  3. Avatar & Banner (Optional): You can leave these fields blank. Portal One will automatically generate an avatar and banner for your agent based on its Name and Description. For this first agent, let's keep it simple and leave them blank.
  4. Click the "Submit" button.

You'll now be taken to the Agent Details page for "Summarizer" landing on the "Config" tab.

Step 3: Configure the Agent's LLM

Your agent needs to be connected to a Large Language Model (LLM) to think and respond.

After creating our agent we are redirected to the agent details page. To configure the model click the 'Config' tab and click the gear icon.
  1. On the "Summarizer" Agent Details page, click on the "Config" tab.
  2. Look for the "Model" card in the first section. Click the gear icon (⚙️) on this card.
  3. Configure the model:
    • Provider: Select the LLM provider you connected earlier (e.g., OpenAI).
    • Model: Choose an available model from the list (e.g., gpt-3.5-turbo or gpt-4.1 if that's the specific default you want to highlight).
    • Advanced Settings (Optional): You'll see an accordion for "Advanced settings." For your first agent, you can safely leave these at their default values.
  4. Click "Submit" button to save the settings and close the dialog.
Select the provider that you configured in the previous section, and choose the model you would like to use.

Step 4: Give Your Agent a Task via Chat

Now, let's interact with "Summarizer."

Clicking on the Chats tab will show a list of all the conversations with the Summarizer agent. Click on the conversation title to drop into the chat pane.
  1. Navigate to the "Chats" tab on the Agent Details page.
  2. An initial chat is created automatically when an agent is made. Click on the title of this initial chat in the table to open the conversation view.
    1. (Alternatively, you could click "+ New Chat" for a fresh conversation.)
  3. You're now in the conversation view (the chat pane).
  4. In the chat input field at the bottom, type your instruction. Your agent's "Description" helps it understand its general role, but specific tasks should be explicitly provided in the chat. Copy and paste the following:
    1. Summarize this text: The quick brown fox jumps over the lazy dog. This classic pangram contains all the letters of the English alphabet. It's often used for testing typewriters and keyboard layouts, as well as for font displays. Understanding how such sentences are constructed can be interesting for language enthusiasts.
  5. Press Enter or click the send button.

If you see the following message make sure to follow the instructions in Connecting Your First LLM Provider.

I'm sorry, I encountered an error and was unable to generate a response.
Error: litellm.AuthenticationError: AuthenticationError: OpenAIException - The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable

Step 5: View the Agent's Response

Watch as "Summarizer" processes your request and streams its summary back to you in the chat pane.

Example of what you might see as a response:

"This text describes the pangram 'The quick brown fox jumps over the lazy dog,' explaining its use in testing keyboards and fonts, and its appeal to language enthusiasts."
After you send the text, 'Summarizer' thinks and then streams its summary back into the chat pane, as shown in this response.

Note on Other Task Methods: While we used chat for this first task, Portal One offers other ways to provide instructions (like "Memories" for permanent instructions), which you can explore later.

Congratulations! You've successfully created your first AI agent, configured its LLM, and had it perform a task.