Integrations
Confident AI supports a wide range of integrations to seamlessly fit into your existing workflow. Whether you're using native SDKs or third-party frameworks, we've got you covered.
Overview
Integrations come in handy for two different workflows:
- When you wish to trace your LLM app, and
- When you wish to run evals in development
All integrations support the former use case. For running evals in development, you'll need to check in each individual integration's documentation pages to know whether they are capable of running end-to-end, component-level, and multi-turn evals.
Connecting this to your stack?Bring traces and evals into the tools your team already usesTalk to an expertOpenTelemetry (OTEL)
OpenTelemetry is an open-source, vendor-neutral observability framework. Confident AI natively accepts OTEL traces at https://otel.confident-ai.com, making it the best option for teams that:
- Use any programming language (not just Python/TypeScript)
- Already have an OTEL-based observability stack
- Want a standards-based approach to tracing with no vendor lock-in
- Need distributed tracing across multiple services
OpenTelemetry Overview
How the OTEL server works, and the gen_ai, confident, OpenInference, and
OpenLLMetry conventions it accepts.
Manual Instrumentation
Quickstarts for raw OpenTelemetry SDKs and the full attribute reference.
Distributed Tracing
Correlate traces across multiple services and microservices.
Third-Party Integrations
Auto-instrument your LLM application with one-line integrations for popular frameworks and providers.
Every confident-trace integration can attach
user and
customer IDs to an auto-instrumented
trace through a trace context.
OpenAI
Chat completion and responses APIs.
LangChain
Framework for building AI applications.
Pydantic AI
Type-safe agent framework for Python.
LangGraph
Graph-based framework for stateful AI applications.
Deep Agents
Agent framework with filesystem tools and subagent delegation.
OpenAI Agents
OpenAI's agent framework for intelligent assistants.
Claude Agent SDK
Anthropic's SDK for building production AI agents.
Vercel AI SDK
The TypeScript AI SDK by Vercel for all AI-based applications.
Crew AI
Multi-agent orchestration for collaborative AI workflows.
LlamaIndex
Data framework for RAG systems and knowledge agents.
Strands Agents
AWS open-source agent framework with native OTel tracing.
LiveKit Agents
Realtime voice and multimodal agents in Python and TypeScript.
Pipecat
Open-source Python framework for realtime voice and multimodal agents.
Google ADK
Google's Agent Development Kit for building multi-agent systems.
smolagents
Lightweight agent framework for Python.
Agno
Framework for building agents, teams, and workflows.
Spring AI
Spring's framework for building AI applications in Java.
Cloud Runtimes
Send OpenTelemetry traces from agents deployed on AWS, Microsoft Azure, Google Cloud, and the Claude Platform to Confident AI. Each guide covers the hosted agent's exporter configuration and the platform-specific setup.
AWS Bedrock AgentCore
Send traces from agents hosted on Amazon Bedrock AgentCore Runtime to Confident AI.
Microsoft Foundry
Send OpenTelemetry traces from Microsoft Foundry hosted agents to Confident AI.
Google Gemini Enterprise
Trace agents on Google Agent Runtime, formerly Vertex AI Agent Engine, with Confident AI.
Claude Platform
Send Claude API and Managed Agents traces to Confident AI, no code changes.
LLM Gateways
If you use an LLM gateway or proxy, these integrations automatically capture traces across all the providers you route through.
LiteLLM
Unified API for 100+ LLM providers.
Portkey
Unified interface for interacting with LLMs.
OpenRouter
Access models from multiple providers through one API.
Bifrost
Route model requests through OpenAI- and Anthropic-compatible APIs.
TrueFoundry
Centralized model access, routing, authentication, and governance.
Last updated on