Botanu SDK provides OpenTelemetry-native run-level cost attribution for AI workflows.
Traditional observability tools trace individual requests. But AI workflows are different — a single business outcome (resolving a support ticket, processing an order) might span multiple LLM calls, retries, tool executions, and data operations across different vendors.
Botanu introduces run-level attribution: a unique run_id that follows your entire workflow, enabling you to answer "How much did this outcome cost?"
- Installation - Install and configure the SDK
- Quick Start - Get up and running in 5 minutes
- Configuration - Configuration options and environment variables
- Run Context - Understanding
run_idand context propagation - Context Propagation - How context flows through your application
- Architecture - SDK design and component overview
- LLM Tracking - Track AI model calls and token usage
- Data Tracking - Track database, storage, and messaging operations
- Outcomes - Record business outcomes for ROI calculation
- Auto-Instrumentation - Automatic instrumentation for common libraries
- Kubernetes Deployment - Zero-code instrumentation at scale
- Existing OTel Setup - Integrate with existing OpenTelemetry deployments
- Collector Configuration - Configure the OpenTelemetry Collector
- Best Practices - Recommended patterns for production use
- Anti-Patterns - Common mistakes to avoid
- Decorators -
@botanu_use_caseand related decorators - Tracking API - Manual tracking context managers
- Configuration API -
BotanuConfigand initialization
from botanu import enable, botanu_use_case
enable(service_name="my-service")
@botanu_use_case(name="my_workflow")
def my_function():
data = db.query(...)
result = llm.complete(...)
return resultApache License 2.0. See LICENSE.