Creating Agents

Follow this step-by-step guide to create a new AI agent in the 8bit-ai dashboard or programmatically via the REST API.

app.8bit-ai.com
Create Agent page

Prerequisites

Before creating an agent, ensure you have:

  • An active 8bit-ai account with email verified
  • An API key if creating agents programmatically
  • An LLM provider configured (OpenAI, Anthropic, etc.)
  • A clear understanding of your agent's purpose and role

Free Tier

The free tier allows up to 3 active agents with GPT-3.5 Turbo. Upgrade to Pro for unlimited agents and access to GPT-4, Claude, and premium models.

Dashboard Creation

The dashboard provides a visual interface for creating and configuring agents.

1

From the dashboard sidebar, click Agents to open the agent management page. Click the Create Agent button in the top right corner.

2

Basic Information

Enter a name for your agent (e.g., "Customer Support Bot") and an optionaldescription that explains the agent's purpose. Choose a modelfrom the available LLM options.

3

System Prompt

Write a system prompt that defines your agent's personality, role, tone, and behavioral guidelines. This is the most critical step — the prompt directly controls response quality.

4

Configure Settings

Adjust LLM parameters like temperature, max tokens, and stop sequences. Optionally enable voice pipeline, connect knowledge bases, and configure channels.

5

Save and Deploy

Review your configuration, click Save to create the agent in draft state, then Deploy to make it live and available for conversations.

API Creation

Create agents programmatically using the REST API for automation and CI/CD workflows.

API Key Required

All API requests require a valid API key passed in the Authorization header. Generate API keys from the Dashboard under Settings → API Keys.

Validation and Testing

After creation, validate your agent's behavior before deploying to production.

Playground Testing

Use the built-in playground in the dashboard to send test messages and review responses. Iterate on your system prompt and parameters based on real interactions.

Edge Case Testing

Test with unexpected inputs, ambiguous questions, and edge cases to ensure your agent handles them gracefully. Verify that sensitive topics are handled appropriately.

Response Quality

Monitor response relevance, accuracy, and tone. Use the session logs to review real user interactions and identify areas for improvement.