Quick Start
Agents
Orchestration
Memory
from pilott import Serve
from pilott.core import AgentConfig, LLMConfig
# Configure LLM
llm_config = LLMConfig(
model_name="gpt-4",
provider="openai",
api_key="your-api-key"
)
async def main():
# Initialize system
pilott = Serve(name="DocumentProcessor")
# Add processing agent
agent = await pilott.add_agent(
role="processor",
goal="Process documents efficiently",
llm_config=llm_config
)
# Execute task
result = await pilott.execute([{
"description": "Extract insights from Q1 report",
"file_path": "documents/q1_report.pdf"
}])
print(f"Processing result: {result[0].output}")
if __name__ == "__main__":
import asyncio
asyncio.run(main())
# Create specialized agents
research_analyst = await pilott.add_agent(
role="research_analyst",
goal="Analyze data and provide insights",
tools=["data_analyzer", "research_synthesizer"],
llm_config=llm_config
)
customer_service = await pilott.add_agent(
role="customer_service",
goal="Handle customer inquiries professionally",
tools=["email_sender", "ticket_manager"],
llm_config=llm_config
)
# Execute agent-specific tasks
analysis_result = await research_analyst.execute_task({
"type": "analyze_market_data",
"data_source": "market_survey_2024",
"variables": ["price", "demand"]
})
# Configure orchestration
from pilott.orchestration import LoadBalancer, DynamicScaling
# Setup load balancing
lb_config = {
"check_interval": 30,
"overload_threshold": 0.8
}
load_balancer = LoadBalancer(pilott, config=lb_config)
await load_balancer.start()
# Configure dynamic scaling
scaling_config = {
"min_agents": 2,
"max_agents": 10,
"scale_up_threshold": 0.8
}
dynamic_scaling = DynamicScaling(pilott, config=scaling_config)
await dynamic_scaling.start()
# System automatically scales based on load
for i in range(100):
pilott.execute([create_task(f"Task {i}")])
# Enhanced memory management
from pilott.memory import EnhancedMemory
# Store and retrieve semantic information
await agent.memory.store_semantic(
text="Important customer feedback on the Q1 product launch",
metadata={"source": "customer_survey", "sentiment": "positive"},
tags={"feedback", "product_launch", "q1"}
)
# Retrieve relevant context
related_info = await agent.memory.semantic_search(
query="product launch feedback",
tags={"customer_survey"},
limit=5
)
# Memory persists across agent executions
await agent.memory.store_task(
task_id="task-123",
task_data={"description": "Handle product inquiry", "status": "completed"}
)
