Skip to main content

Build Your First Agent

In this tutorial, you'll build a simple customer inquiry classifier agent. By the end, you'll understand the core Waxell concepts.

Prerequisites​

  • Python 3.10+
  • Waxell SDK and Runtime installed
pip install waxell-sdk waxell-runtime

Step 1: Define the Agent​

Create a new file inquiry_agent.py:

from waxell_sdk import agent, workflow, decision

@agent(
    name="inquiry-classifier",
    description="Classifies customer inquiries"
)
class InquiryAgent:
    pass

The @agent decorator defines a container for your agent's capabilities.

Step 2: Add a Decision​

Decisions are LLM-powered classification or generation points:

@agent(name="inquiry-classifier")
class InquiryAgent:

    @decision
    def classify(self, ctx):
        """Classify the inquiry type."""
        return ctx.llm.classify(
            ctx.input.message,
            categories=["billing", "technical", "general", "complaint"]
        )

Step 3: Add a Workflow​

Workflows orchestrate decisions and actions:

@agent(name="inquiry-classifier")
class InquiryAgent:

    @workflow
    def process_inquiry(self, ctx):
        # Classify the inquiry
        category = ctx.call(self.classify)

        # Return structured result
        return {
            "category": category,
            "message": ctx.input.message,
            "confidence": "high"
        }

    @decision
    def classify(self, ctx):
        return ctx.llm.classify(
            ctx.input.message,
            categories=["billing", "technical", "general", "complaint"]
        )

Step 4: Run the Agent​

from waxell_runtime import RuntimeConfig

# Get runtime configuration
config = RuntimeConfig.get()

# Create agent instance
agent = InquiryAgent()

# Run the workflow
result = config.run(
    agent.process_inquiry,
    input={"message": "I can't log into my account"}
)

print(result)
# {"category": "technical", "message": "I can't log into my account", "confidence": "high"}

Complete Example​

from waxell_sdk import agent, workflow, decision
from waxell_runtime import RuntimeConfig

@agent(
    name="inquiry-classifier",
    description="Classifies customer inquiries",
    version="1.0.0"
)
class InquiryAgent:

    @workflow
    def process_inquiry(self, ctx):
        category = ctx.call(self.classify)
        return {
            "category": category,
            "message": ctx.input.message
        }

    @decision
    def classify(self, ctx):
        return ctx.llm.classify(
            ctx.input.message,
            categories=["billing", "technical", "general", "complaint"]
        )

if __name__ == "__main__":
    config = RuntimeConfig.get()
    agent = InquiryAgent()

    result = config.run(
        agent.process_inquiry,
        input={"message": "My payment didn't go through"}
    )
    print(f"Classification: {result['category']}")

Next Steps​