Strands (Python)
AWS-backed framework with native OTLP support.
Integrations
SideSeat collects runs from any application emitting OpenTelemetry data. Most AI frameworks have built-in OpenTelemetry support or can be instrumented with the SideSeat SDKs.
How Integration Works
Section titled “How Integration Works”- Install SDK: Add the SideSeat Python or TypeScript SDK
- Initialize: Call
SideSeat()(Python) orinit()(TypeScript) - Run: Your framework’s telemetry flows to the local workbench automatically
SideSeat then:
- Auto-detects the framework
- Extracts model, token, and cost information
- Normalizes messages into a unified run timeline
Frameworks
Section titled “Frameworks”AI agent frameworks with OpenTelemetry support:
Strands (TypeScript)
TypeScript SDK for Strands agents on Bedrock.
Vercel AI SDK
TypeScript AI toolkit for generateText/streamText.
Google ADK
Google’s Agent Development Kit for Gemini-powered agents.
LangGraph
LangChain’s agent framework for stateful workflows.
OpenAI Agents SDK
OpenAI’s agent framework with Logfire instrumentation.
Microsoft Agent Framework
Microsoft’s agent SDK with OpenAI and Anthropic support.
CrewAI
Multi-agent framework for collaborative AI.
Other Frameworks
LangChain, PydanticAI, LlamaIndex and generic OpenTelemetry.
Providers
Section titled “Providers”LLM providers from which SideSeat extracts model information:
Amazon Bedrock
Multi-model platform on AWS.
Anthropic
Claude models via API.
OpenAI
GPT models via API.
Azure OpenAI
Google Gemini
Gemini models via Google Gen AI SDK.
Google Vertex AI
Google Cloud AI platform via the native Vertex AI SDK.