Torale DocsTechnical Documentation
Grounded search monitoring platform
Skip to content
Overview
Torale executes scheduled web searches, evaluates results against trigger conditions, and stores execution history. Grounded search combines Google Search with LLM evaluation for source-backed monitoring.
Example
python
from torale import ToraleClient
client = ToraleClient()
# Create a monitoring task
task = client.tasks.create(
search_query="When is the iPhone 17 being released?",
condition_description="Apple has announced a specific release date",
schedule="0 9 * * *" # Daily at 9 AM
)
# Check results
executions = client.tasks.get_executions(task.id)
if executions[0].condition_met:
print(executions[0].result["answer"])Architecture
Core Components:
- Monitoring Agent: Gemini-powered agent (Pydantic AI) with Perplexity search and Mem0 memory
- APScheduler: Cron-based scheduling with agent-driven dynamic rescheduling
- Condition Evaluation: Agent determines if trigger condition is met, returns evidence + sources
- Notifications: Email and webhook delivery when conditions are met
Interfaces
- Web Dashboard - Browser-based task management
- CLI - Command-line interface
- Python SDK - Programmatic access
- API - REST API reference
- Self-Hosted - Deploy on your infrastructure