Few-shot prompting is a powerful technique where you provide a small number of sample interactions (typically 2 to 5) to guide your agent’s behavior. This method helps the agent understand the desired output format and task requirements by learning from the given examples, thereby improving performance without writing extensive instructions.

​
Crafting Effective Examples

  • Provide Task Demonstrations: Use examples that clearly illustrate the tasks that your agents will perform.
  • Use Realistic Scenarios: Include interactions that mirror actual conversations that your agent will handle.
  • Use Preferred Tone and Style: Ensure the agent’s replies in your examples match your desired brand voice.

​
Defining Few-Shot Examples

In the Agency Swarm framework, few-shot examples are structured using the OpenAI message object format, including the role and content fields.

Example: Technical Support Agent:

examples = [
    {
        "role": "user",
        "content": "My device won't turn on.",
    },
    {
        "role": "assistant",
        "content": "I'm sorry to hear that. Let's try some troubleshooting steps. First, please press and hold the power button for at least 10 seconds.",
    },
    {
        "role": "user",
        "content": "I tried that, but it still won't turn on.",
    },
    {
        "role": "assistant",
        "content": "Thank you for trying that. Please connect your device to a charger and check if any lights appear. Let me know what you observe.",
    }
]

The optional fields attachments and metadata can be included if needed but are not required for basic examples.

​
Using Few-Shot Examples

You can add few-shot examples to your agent either during initialization or afterward:

from agency_swarm import Agent

agent = Agent(
    name="CustomerSupportAgent",
    description="Assists customers with inquiries and provides detailed information.",
    examples=examples
)

See more advanced features in Agent Class